Selective cyclooxygenase-2 inhibition suppresses basic fibroblast growth factor expression in human esophageal adenocarcinoma

MOLECULAR CARCINOGENESIS 46:971–980 (2007)

Selective Cyclooxygenase-2 Inhibition Suppresses Basic Fibroblast Growth Factor Expression in Human Esophageal Adenocarcinoma Mark Baguma-Nibasheka,1 Christie Barclay,1 Audrey W. Li,1 Laurette Geldenhuys,2 Geoffrey A. Porter,3,4 Jonathan Blay,5 Alan G. Casson,2,3,6 and Paul R. Murphy1* 1

Department of Physiology and Biophysics, Faculty of Medicine, Dalhousie University, Halifax, Nova Scotia, Canada Department of Pathology, Faculty of Medicine, Dalhousie University, Halifax, Nova Scotia, Canada 3 Department of Surgery, Faculty of Medicine, Dalhousie University, Halifax, Nova Scotia, Canada 4 Department of Community Health and Epidemiology, Faculty of Medicine, Dalhousie University, Halifax, Nova Scotia, Canada 5 Department of Pharmacology, Faculty of Medicine, Dalhousie University, Halifax, Nova Scotia, Canada 6 Department of Surgery, University of Saskatchewan, Royal University Hospital, Saskatoon, Saskatchewan, Canada 2

Inhibition of cyclooxygenase (COX)-2 is reported to suppress growth and induce apoptosis in human esophageal adenocarcinoma (EADC) cells, although the precise biologic mechanism is unclear. In this study we tested the hypothesis that the antitumor activity of COX-2 inhibitors may involve modulation of basic fibroblast growth factor (FGF-2), which is overexpressed in EADC. We evaluated the effects of NS-398, a selective COX-2 inhibitor, on FGF-2 expression and proliferation of EADC cell lines that express COX-2 and those that do not. We also correlated COX-2 and FGF-2 expression with clinico-pathologic findings and outcome in a well-characterized series of surgically resected EADC tissues. Seg-1 cells robustly expressed COX-2 and FGF-2, whereas Bic-1 cells expressed neither transcript. FGF-2 was reduced to undetectable levels in Seg-1 cells following NS-398 treatment, but increased within 4 h of drug removal. NS-398 significantly inhibited the growth of Seg-1 cells, and this effect was ameliorated by addition of exogenous FGF-2. In contrast, NS-398 had no effect on Bic-1 cell proliferation and FGF-2 alone had no effect on proliferation of either cell line. NS-398, or a neutralizing anti-FGF-2 antibody, induced apoptosis in Seg-1 cells, and these effects were inhibited by addition of exogenous FGF-2. COX-2 protein was strongly expressed in 46% (10/22) of EADCs, and was associated with a trend towards reduced disease-free survival. These findings indicate that the antitumor effects of COX-2 inhibition in EADC cells may be mediated via suppression of FGF-2, and that COX-2 may be a clinically relevant molecular marker in the management of human EADC. ß 2007 Wiley-Liss, Inc. Key words: esophageal adenocarcinoma; COX-2 inhibitor; FGF-2; anti-FGF

INTRODUCTION Over the past three decades, a marked change in the epidemiology of esophageal malignancy in North America and Europe has been reported, with an increasing incidence of esophageal adenocarcinoma (EADC) [1,2]. Primary EADC is thought to arise from Barrett esophagus (BE), an acquired condition predisposed by gastroesophageal reflux disease (GERD), in which the normal esophageal squamous epithelium is replaced by a specialized metaplastic columnar cell epithelium [3]. Progression of BE to invasive EADC is reflected histologically by the metaplasia–dysplasia–carcinoma sequence. The recent identification of molecular markers associated with BE and EADC may provide further insight not only into the molecular pathogenesis of this disease, but towards the development of novel therapeutic and chemopreventive strategies [4,5]. Basic fibroblast growth factor (FGF-2) is the prototypic member of a family of related genes encoding ß 2007 WILEY-LISS, INC.

heparin-binding proteins with growth-, anti-apoptotic- and angiogenic activity [6]. FGF-2 is overexpressed in human esophageal cancer cell lines and in primary tumor tissues, implicating an autocrine or paracrine role for FGF-2 in esophageal tumorigenesis [7,8]. We recently characterized the expression of FGF-2 and its natural endogenous antisense RNA (FGF-AS/GFG) in human esophageal cancer cell lines and surgically resected esophageal tissues, and reported that overexpression of FGF-2 mRNA is a significant predictor of tumor recurrence and mortality in patients after esophageal cancer resection [9].

*Correspondence to: Department of Physiology and Biophysics, Faculty of Medicine, Dalhousie University, 5850 College Street, Halifax, NS B3H 1X5, Canada. Received 23 January 2007; Revised 12 March 2007; Accepted 23 March 2007 DOI 10.1002/mc.20339

972

BAGUMA-NIBASHEKA ET AL.

Furthermore, these risks were ameliorated in tumors co-expressing FGF-AS/GFG, supporting our hypothesis that GFG is novel tumor suppressor modulating the effects of FGF-2 expression [9]. These novel observations suggest that FGF-2 expression is an important prognostic indicator in EADC, and identify FGF-2 as a potential molecular target in the treatment of this disease. One possible mechanism for regulation of FGF-2 expression and/or action is via cyclooxygenase (COX), a key enzyme in the synthesis of prostaglandins. There are at least two distinct COX isoenzymes: COX-1, which is constitutively expressed in many normal tissues and is responsible for various physiologic functions, and COX-2, an inducible prostaglandin synthase upregulated in gastrointestinal tumors in response to inflammation [10]. Several recent studies have reported upregulation of COX-2 in both BE and EADC tissues, suggesting a functional role in Barrett metaplasia–dysplasia–adenocarcinoma progression [11–17]. Recent experimental [18] and epidemiological [19] evidence suggests that long-term use of COX-2 inhibitors may reduce the incidence of esophageal cancer, and clinical trials are ongoing to assess the chemopreventive efficacy of COX-2 inhibition in patients [20]. The selective COX-2 inhibitors N-[2(cyclohexyloxy)-4-nitrophenyl]-methanosulfonamide (NS-398) and rofecoxib have been shown to inhibit esophageal cell proliferation and to induce apoptosis in COX-2 expressing esophageal cell lines [21–24]. Although the precise biological mechanisms underlying these observations are unclear, it is possible that the effect of COX-2 inhibition may be mediated, at least in part, via suppression of FGF-2 expression [25–27] or action [28,29]. Suppression of angiogenesis and growth of gastrointestinal tumor explants in nude mice following administration of oral COX-2 inhibitors was reported to be associated with reduced expression of FGF-2 and VEGF [25]. Furthermore, in a rodent model to assess healing of gastric ulcers, treatment with rofecoxib significantly reduced expression of FGF-2 and delayed ulcer healing, whereas no effect was seen on VEGF expression [26,27]. To test the hypothesis that the antitumor activity of COX-2 inhibitors may be mediated through suppression of FGF-2 expression or activity, the primary objective of this study was to investigate the interaction of FGF-2 and the highly selective COX-2 inhibitor NS-398 in Barrett-associated EADC cell lines that express FGF-2 and COX-2 (Seg-1) and those that do not (Bic-1). Anticipating potential clinical application of COX-2 inhibition in the management of human EADC, a secondary objective of this study was to characterize COX-2 (mRNA and protein) expression in a well defined series of surgically resected EADC tissues (already characterized for FGF-2 expression [9]), and to correlate Molecular Carcinogenesis DOI 10.1002/mc

expression with clinico-pathologic findings and outcome. MATERIALS AND METHODS Chemicals and Reagents Culture medium, fetal bovine serum (FBS) and all other culture reagents were from InVitrogen (Burlington, Ont., Canada). Gels, membranes and other materials used in Western blotting were, unless otherwise noted, from Bio-Rad Laboratories, Hercules, CA. Rabbit polyclonal anti-FGF-2 (Ab-2) was from CalBioChem, La Jolla, CA. Alexa FluorTM 488 (green) goat anti-rabbit fluorescent IgG conjugate from InVitrogen and donkey anti-rabbit Ig-horseradish peroxidase conjugate from GE HealthCare Biosciences, Piscatawy, NJ. All other chemicals were from Sigma Chemical Company, St. Louis, MO, unless otherwise specified. Cell Lines and Culture Two human Barrett-associated EADC cell lines (Seg-1 and Bic-1) were generously provided by Dr. David G. Beer, University of Michigan, Ann Arbor, MI. Cells were grown as monolayers in Dulbecco’s modified Eagle’s medium (DMEM) with L-glutamine, supplemented with 10% FBS, penicillin at 100 units/ml and streptomycin (0.1 mg/ml) and kept at 378C in a humidified, 5% CO2 chamber. RNA Isolation and RT-PCR Amplification Total RNA was isolated using the RNeasyTM kit from Qiagen, Inc., Mississauga, Ont., Canada, according to the manufacturer’s instructions. Then, following reverse transcription with M-MLV reverse transcriptase (Promega, Madison, WI), amplification of the FGF-2 and COX-2 transcripts was as previously described [9,21]. Briefly, thermal cycling consisted of: (i) 30 cycles at 948C for 1 min, 638C for 45 s and 728C for 1.5 min with the primers 50 -GGCTTCTTCCTGCGCATCCA-30 (forward) and 50 -GCTCTTAGCAGACATTGGAAGA-30 (reverse) for the 352 bp FGF2 product or (ii) 35 cycles at 948C for 1 min, 598C for 1 min and 728C for 1.5 min with the primers 50 CAGCACTTCACGCATCAGTT-30 (forward) and 50 TCTGGTCAATGGAAGCCTGT-30 (reverse) for the 756 bp COX-2 product. FGF-2 and COX-2 levels were normalized against the glyceraldehyde 3-phosphate dehydrogenase (GAPDH) transcript amplified from the same RT reaction. Extraction and Western Analysis of FGF-2 Monolayer cell cultures were washed twice with phosphate-buffered saline (PBS), scraped into PBS and isolated by centrifugation (300g, 3 min at 48C). Cells were resuspended in ice-cold lysis buffer (500 ml of 50 mM Tris–HCl, pH 7.4, 150 mM NaCl, 1 mM EGTA and 50 mM sodium pyrophosphate, containing 3 mg/ml each of aprotinin, leupeptin, pepstatin

COX-2 AND FGF-2 IN ESOPHAGEAL ADENOCARCINOMA

and 2 mM phenylmethylsulfonyl fluoride) for 30 min, then homogenized by gentle passage 10 times through a 21G needle. After 30 min on ice, the homogenates were centrifuged (10 000g, 20 min at 48C) and the supernatant taken as the total cell lysate. Protein concentration was measured with the Coomassie assay (Pierce, Rockford, IL), and equal amounts of protein (40 mg) from each lysate were solubilized in 3
SDS-sample buffer, separated on 15% SDS-PAGE gels and transferred to nitrocellulose for Western analysis as previously described [30,31]. Rabbit polyclonal anti-FGF-2 and donkey anti-rabbit Ig-horseradish peroxidase conjugate were used at 2 mg/ml and 0.1 mg/ml, respectively, in Blotto (1
Tris-buffered saline, 10% skim milk, 0.05% Tween-20). Immunoreactive bands were detected with chemiluminescence (SuperSignalTM, Pierce). Immunofluorescent Confocal Laser Scanning Microscopy For immunofluorescence studies, cells at approximately 25% confluence in culture flasks were trypsinized and re-seeded into dishes of equal area containing microscope slides and allowed to attach for 18 h before further manipulation. Following appropriate incubation, the cells were fixed with icecold 2% paraformaldehyde (with 9 mg/ml disodium hydrogen orthophosphate and 6 mg/ml L-lysine) for 15 min, permeabilized with cold 0.1% Triton X-100 in PBS for 15 min and then blocked with 3% bovine serum albumin (BSA) in PBS for 1 h [32]. Staining was by sequential exposure to the primary antibody and the fluorescently tagged secondary antibody, each for an hour at room temperature. All antibodies were diluted in 0.1% BSA-PBS as follows: anti-FGF-2, 10 mg/ml; Alexa FluorTM 488 (green) anti-rabbit, 40 mg/ml. Slides were subsequently stained for DNA using propidium iodide (PI, 50 ng/ml in PBS) for 10 min, then mounted in a drop of glycerol-PBS (CitifluorTM, Marivac, Halifax, NS, Canada). Image analysis used the standard operating software on the Zeiss LSM 510 microscope. Proliferation and Cell Cycle Analyses Cells were plated on culture dishes at a density of 2
105 cells/cm2 in DMEM and allowed to attach overnight. The media were then changed to DMEM (control) or DMEM with 3.3 mg/ml (10 mM) NS-398 (Cayman Chemicals, Ann Arbor, MI), 10 ng/ml recombinant human FGF-2 (EMD Biosciences, San Diego, CA) or both. In the proliferation assay, the cells were harvested by trypsinization on successive days and counted through a Coulter counter. For cell cycle analysis, the trypsinized cells were centrifuged (400g, 10 min at 48C), fixed overnight at 48C in 70% ethanol and incubated for 30 min at room temperature in PBS with 50 mg/ml PI and 100 units/ml ribonuclease A. Fluorescence intensities were determined in a Becton Dickinson FACSCalibur flow cytometer and the proportion of cells in different Molecular Carcinogenesis DOI 10.1002/mc

973

phases of the cycle estimated using the ModFit program. Apoptosis Assay To investigate the effects of NS-398 and FGF-2 on apoptosis, the FAM FLICATM Poly-Caspases Assay Kit (Immunochemistry Technologies, Bloomington, MN) was used according to the manufacturer’s protocol. Briefly, cells were incubated with DMEM (control) or DMEM with 3.3 mg/ml NS-398, 10 ng/ml FGF-2 or both for 48 h. The FLICA staining reagent was then added and the cells incubated for a further 2 h before harvest for fluorescence-activated cell sorting (FACS). The proportion of cells immunoreactive for caspases was assessed using FCS Express V2. Data from the four treatment groups were compared using the ANOVA with Student–Newman–Keuls procedure, with differences of P < 0.05 considered significant. Neutralization of FGF-2 Cells growing on culture dishes were washed twice with PBS and the media changed to DMEM with 1% FBS (control) or DMEM with 1% FBS and 4 mg/ml anti-FGF-2 antibody (R&D Systems, Minneapolis, MN), 0.5 ng/ml FGF-2 or both. The cells were then harvested on successive days and assessed for viability/proliferation and apoptosis as described above. Patients and Tumors All participating patients gave informed, written consent. Collection and storage of resected esophageal tissues was in accordance with the 1998 Canadian Tri-Council Policy ‘‘Statement on Ethical Conduct for Research Involving Humans’’. Approval to study banked esophageal tissues was approved by the Capital Health Research Ethics Board (QE-2000277) and the Health Sciences and Humanities Research Ethics Board at Dalhousie University (2002-539). The study population, including preoperative investigations, surgical approach, and postoperative follow-up was described in detail previously [9]. Here we studied a subset of 22 patients with primary EADC, defined according to strict clinico-pathologic criteria, for whom tissues (tumor and matched histologically normal esophageal epithelium, stored in liquid nitrogen at 808C) were available from our esophageal tumor bank, each with correlative clinical, histopathologic, staging and outcomes data. Follow-up was complete for all patients until July 2006. FGF-2 and COX-2 Expression in Human EADC Tissues RT-PCR was used to study FGF-2 and COX-2 mRNA expression (normalized to GAPDH) in tumors and matched histologically normal esophageal epithelia (internal control), as described [9]. Levels of FGF-2 and COX-2 mRNA expression in tumors, relative to

974

BAGUMA-NIBASHEKA ET AL.

matched normal epithelia, were stratified initially as follows: underexpressed (ratio below 0.7); the same (0.71–1.5); or overexpressed (above 1.5). All PCR products were obtained within the linear range of the reaction. All RT-PCR assays were performed (at minimum) in duplicate on coded samples by a graduate student without knowledge of clinical correlative and outcome data. A modified indirect immunoperoxidase assay was used to study FGF-2 and COX-2 protein expression and distribution in serial unstained formalin-fixed, paraffin-embedded tissue sections (4 mm thickness), as described [9]. Primary antibodies included affinitypurified anti-FGF-2 polyclonal antibodies raised against a synthetic peptide corresponding to amino acids 40–63 of human FGF-2 (CalBioChem) and a goat polyclonal IgG antibody against COX-2 (M-19; Santa Cruz Biotechnology, Santa Cruz CA). Briefly, sections were incubated overnight at 48C with primary antibodies against FGF-2 at 1:50 dilution and COX-2 (1.3 mg/ml) in a high-humidity chamber. Subsequent steps were performed using Universal LSAB plus and DAB plus kits according to the manufacturer’s protocols (DAKO Corp., Carpenteria, CA). Laboratory controls were run in parallel with test sections and included known positive and negative tissues (tissue controls), and sections stained without the primary antibody (reagent controls). Interpretation of coded tissue sections was performed independently by two investigators blinded to mRNA expression or associated clinicopathologic data, and consensus reached at a doubleheaded microscope. To overcome the issue of tissue heterogeneity and to minimize subjectivity, a semiquantitative, validated composite score, based on intensity of immunoreactivity (0, no staining; 1, weak; 2, intermediate; 3, strong), and proportion of immunopositive cells (0, none; 1, less than one hundredth; 2, one hundredth to one tenth; 3, one tenth to one third; 4, one third to two thirds; 5, greater than two thirds) was assigned to each tissue section, as reported [9]. Overall accumulation of FGF2 or COX-2 protein was then expressed as the sum of the intensity and proportion scores (range: 0, and 2– 8). Tissues were considered negative with a composite score of 0, 2 or 3 (thereby avoiding a false positive result from occasional immunopositive cells). Tissues were considered to have weak positivity with a composite score of 4–6 (low stringency), and to have strong positivity with a composite score of 7 or 8 (high stringency). The subcellular distribution of each protein (cytoplasmic, nuclear, or both) was recorded.

stage), were tested with a Chi-square test, with a Fisher exact test used if a cell contained fewer than five patients. The prognostic importance of FGF-2 and COX-2 mRNA expression (underexpressed vs. same vs. overexpressed; underexpressed/same vs. overexpressed; underexpressed vs. same/overexpressed) and protein expression (negative vs. weak positivity vs. strong positivity) in tumors for overall and disease-free survival was examined in a univariate analysis with Kaplan–Meier survival methods and tested with the log-rank test. Multivariate analysis using Cox proportional hazards was then used to adjust for the effects of age, gender, tumor grade, pT-stage, pN-stage and overall UICC stage. Statistical significance was set at P ¼ 0.05 and all analyses were performed using SPSS for Windows 13.0 (SPSS, Inc., Chicago, IL). RESULTS Expression of FGF-2 and COX-2 in EADC Cell Lines Using RT-PCR to examine the expression of FGF-2 mRNA in EADC cells, we first demonstrated that FGF2 was expressed in Seg-1, but not in Bic-1 cell lines (Figure 1), in keeping with our initial observation [9]. As differential expression of the COX-2 transcript was previously described in these EADC cell lines [21], we confirmed that Seg-1 cells expressed a high level of COX-2 mRNA, whereas COX-2 was not expressed in Bic-1 cells (Figure 1). Effect of NS-398 on FGF-2 Expression Consistent with the mRNA data, immunoreactive FGF-2 was abundant in Seg-1 cells, but was negligible in Bic-1 (Figure 2A). Incubation with the selective COX-2 inhibitor, NS-398 (3.3 mg/ml for 48 h) resulted in a marked suppression of cellular FGF-2 content in

Statistical Analysis Differences in the frequency of FGF-2 and COX-2 mRNA and protein expression, according to demographic and clinico-pathologic factors (age, gender, tumor differentiation, pT stage, pN stage, UICC Molecular Carcinogenesis DOI 10.1002/mc

Figure 1. Expression of FGF-2 and COX-2 in human EADC cell lines. Representative RT-PCR results illustrate the relative expression of FGF-2 and COX-2 mRNA transcripts in Seg-1 and Bic-1 EADC cells. The level of GAPDH (glyceraldehyde 3-phosphate dehydrogenase) in each sample was used as an internal loading control.

COX-2 AND FGF-2 IN ESOPHAGEAL ADENOCARCINOMA

975

Figure 2. Effect of COX-2 inhibition on FGF-2 expression in EADC cells. (A) Immunofluorescent detection of FGF-2 in Seg-1 (upper panel) and Bic-1 (lower panel) cells before and after treatment with NS-398. Cells were incubated without (CTR) or with 3.3 mg/ml NS-398 for 48 h and stained for FGF-2 (green) and DNA (red) at the indicated times after NS-398 removal. (B) Western blot detection of FGF-2 in Seg-1 cells before (CTR) and after treatment with NS-398 as in (A). A high molecular weight non-specific band (ns) is shown as a loading control.

Seg-1 cells as indicated by fluorescent immunohistochemistry or Western blotting (Figure 2A and B) and by flow cytometry (data not shown). However, FGF-2 levels increased rapidly within 4 h of NS-398 washout and return to normal medium. NS-398 did not have a detectable effect on FGF-2 protein levels in Bic-1 cells (Figure 2A, lower panels). Effect of NS-398 and FGF-2 on Cell Growth NS-398 significantly inhibited the growth of Seg-1 cells (7.80  0.1
105 total cell count, mean  SEM, at day 4 in 3.3 mg/ml NS-398 vs. 28.1  1.4
105 for cells in normal medium), and this effect was preventable by the presence of exogenous FGF-2 (10 ng/ml) in the culture medium with NS-398 (total cell count 23.33  0.6
105, Figure 3A). In contrast, NS-398 had no effect on the growth of Bic-1 cells (Figure 3B), and exogenous FGF-2 alone had no significant effect on either cell line. Effect of NS-398 and FGF-2 on Cell Cycle Progression and Apoptosis The ability of FGF-2 to reverse the inhibitory effect of NS-398 on cell proliferation could be attributable to stimulation of cell cycle progression or inhibition of apoptosis. NS-398 treatment for 48 h induced a significant fourfold increase in the apoptotic index, from 4.3  0.8% observed in control cultures to Molecular Carcinogenesis DOI 10.1002/mc

Figure 3. Effect of NS-398 (3.3 mg/ml) and FGF-2 (10 ng/ml) on proliferation of Seg-1 (A) and Bic-1 cells (B). Results are the means  SEM of triplicates.

18.32  2.7% in cultures treated with NS-398 for 48 h (Figure 4). FGF-2 alone had no influence on the level of apoptosis. However, FGF-2 significantly reduced NS-398-induced apoptosis from 18.32  2.7% in cultures receiving NS-398 alone to 9.88  1.5% in cultures receiving NS-398 and FGF-2. Treatment with NS-398 for 36 h also resulted in a significant reduction in the percentage of cells in S-phase (10.48  1.6%) compared to 27.33  2.4% in untreated controls (data not shown). In contrast to its effects on apoptosis, exogenous FGF-2 alone, or in combination with NS-398 had no significant effect on the percentage of cells in S-phase. Effect of FGF-2 Neutralizing Antibody In order to determine the role of endogenous FGF-2 in EADC cells, we next examined the effect of a

976

BAGUMA-NIBASHEKA ET AL.

Figure 4. Effects of NS-398 and FGF-2 on apoptosis in Seg-1 cells. (A) Representative fluorescence histogram of caspase staining in Seg-1 cells incubated with 3.3 mg/ml NS-398 for 48 h compared with control cells growing in normal medium. The apoptotic index is derived from the percentage of cells in the indicated region. (B) Apoptosis in Seg-1 cells in normal DMEM (CTR) or in the presence of NS-398, FGF-2 or both. The results represent the mean  SEM of four independent experiments. Groups with different letters are significantly different from each other (P < 0.05, n ¼ 4).

neutralizing FGF-2 antibody on cell proliferation and apoptosis in Seg-1 and Bic-1 cells. Flow cytometry indicated that treatment with the anti-FGF-2 antibody significantly increased apoptosis in Seg-1 cells, raising the apoptotic index from 10.58  1.2% in control cells to 25.79  1.6% within 48 h (Figure 5A). In contrast, the antibody had no effect on apoptosis in Bic-1 cells, which do not express FGF-2 (Figure 5B). Furthermore addition of exogenous FGF-2 (0.5 ng/ ml) significantly reduced apoptosis in antibody treated Seg-1 cultures (Figure 5A). Although FGF-2 alone did not significantly affect apoptosis in Seg-1 cells, it caused a small but significant reduction in apoptosis in Bic-1 cells compared to cells treated with FGF neutralizing antibody. Molecular Carcinogenesis DOI 10.1002/mc

Figure 5. Neutralization of endogenous FGF-2 activity enhances apoptosis in Seg-1 (A) but not Bic-1 cells (B). Cells were cultured for the indicated times in DMEM containing 1% serum without further addition (CTR), or with a neutralizing FGF-2 antibody (FGF-Ab), FGF-2, or both. The apoptotic index (mean  SEM of four separate experiments) was determined as described in the legend to Figure 4. Groups with different letters are significantly different from each other (P < 0.05, n ¼ 4). The asterisk (*) indicates that FGF-2 treated Bic-1 are significantly different from FGF-Ab treated cultures (P < 0.05, n ¼ 4), but not from control cultures.

Associations Between FGF-2 and COX-2 (mRNA and Protein) Expression, Clinico-Pathologic Findings and Outcome in Human EADC The patient series from which pathological tissue specimens were obtained comprised 21 males and 1 female, ranging in age from 51 to 79 yr (median, 62 yr). Five tumors (23%) were well differentiated, 7 (32%) moderately differentiated, and 10 (45%) poorly differentiated. All tumors were pT3, and 15 (68%) had regional lymph node metastases (pN1) which were associated with significantly reduced postoperative survival (median 10 mo vs. greater than 49 mo if pN0; P ¼ 0.02).

COX-2 AND FGF-2 IN ESOPHAGEAL ADENOCARCINOMA

977

Using criteria defined a priori to stratify levels of mRNA expression in tumors (normalized to GAPDH and relative to matched histologically normal esophageal squamous epithelium), FGF-2 mRNA was found to be overexpressed in 36% (8/22) of EADC, comparable to our initial report of 35% FGF-2 mRNA overexpression in a larger series of patients [9]. FGF-2 protein was not detected in any normal esophageal squamous epithelia, but was overexpressed in 86% (19/22) of tumors (stringent criteria), where immunopositivity was localized exclusively to the cytoplasm. Further in keeping with our initial report [9], overexpression of FGF-2 mRNA in this subset of 22 patients with EADC was also associated with reduced postoperative overall and disease-free survival (median 4 mo vs. 33 mo if FGF-2 mRNA was the same/underexpressed; Figure 6C), although this did not reach formal statistical significance (P ¼ 0.06), likely reflecting fewer patient numbers in this study.

COX-2 mRNA was overexpressed in 55% (12/22) of tumors, but no statistically significant association was found between COX-2 mRNA expression and FGF-2 expression (mRNA and/or protein), level of COX-2 protein expression, any clinico-pathologic parameter (age, gender, tumor differentiation, pT stage, pN stage, UICC stage), overall or disease-free survival. COX-2 protein was weakly expressed in the parabasal layer of normal squamous epithelium (considered negative, as the maximum composite immunohistochemical score was 3) and in six tumors (each with composite immunohistochemical score of only 4). Using stringent criteria (a composite immunohistochemical score of 7 or 8), COX-2 protein, which was localized to the cytoplasm of tumor cells (Figure 6B), was found to be strongly expressed in 46% (10/22) of EADCs. Although strong expression of COX-2 protein was associated with a trend towards reduced overall and disease-free

Figure 6. Top panels (A and B): Immunohistochemistry was used to study tissue and subcellular distribution of COX-2 protein in surgically resected esophageal tissues using a goat polyclonal IgG antibody (M-19; 1.3 mg/ml; Santa Cruz Biotechnology). (A) A negative control. (B) Strong cytoplasmic positivity for COX-2 protein is illustrated in a moderately differentiated esophageal adenocarcinoma (400
). Lower panels (C and D): Kaplan–Meier disease-free survival curves for patients after surgical resection of esophageal

adenocarcinoma, categorized by levels of expression of FGF-2 mRNA (C) and COX-2 protein (D). (C) Reduced disease-free survival was seen for patients (8/22) with tumors overexpressing FGF-2 mRNA (solid line) although this did not reach formal statistical significance (P ¼ 0.06). (D) A trend towards reduced disease-free survival was seen for patients (10/22) with tumors strongly immunopositive for COX-2 protein (broken line), although this was not statistically significant (P ¼ 0.09).

Molecular Carcinogenesis DOI 10.1002/mc

978

BAGUMA-NIBASHEKA ET AL.

survival (median 12 mo vs. 27 mo if COX-2 protein was immunonegative or weakly expressed; Figure 6D), this did not reach formal statistical significance (P ¼ 0.09), again most likely reflecting limited patient numbers. DISCUSSION The human EADC cell lines Seg-1 and Bic-1 offer a unique model to investigate the role of FGF-2 and COX-2 in esophageal adenocarcinoma. We have shown that Seg-1 expresses both COX-2 and FGF-2 mRNAs, whereas Bic-1 expresses neither transcript. We have previously reported that FGF-2 expression is a significant predictor of tumor recurrence and mortality in a subset of patients with esophageal cancer. In the present study we investigated the possible regulation of FGF-2 by COX-2, and the role of FGF-2 in survival and proliferation of these two human esophageal EADC cell lines. We have demonstrated that the COX-2 inhibitor NS-398 (which has greater selectivity than either celecoxib or rofecoxib [33])has no effect on the COX-2-negative Bic-1 cells, but significantly inhibited cell proliferation and stimulated apoptosis in Seg-1 cells. This was accompanied by a profound but reversible suppression of FGF-2 expression in Seg-1 cells. Furthermore, addition of exogenous FGF-2 to the culture medium significantly ameliorated the anti-proliferative and pro-apoptotic effects of NS-398 in Seg-1. In contrast, compared to controls, FGF-2 had no significant effect on proliferation or apoptosis of Bic-1 cells, in the presence or absence of NS-398. Taken together, these data suggest that in Seg-1 adenocarcinoma cells, COX-2 promotes cell survival and proliferation, at least in part, via an FGF-2 dependent pathway. Two human cyclooxygenase genes (COX-1 and COX-2) have been identified, and both encode enzymes involved in prostaglandin synthesis. COX-1 is constitutively expressed in many tissues, whereas COX-2 is induced in response to stimuli including mitogens, inflammatory cytokines, and tissue injury. Both COX-1 and COX-2 are inhibited to varying degrees by all of the available nonsteroidal anti-inflammatory drugs [34]. However, the inhibition of proliferation and induction of apoptosis observed in the present study was elicited with 10 mM NS-398, a concentration at which it is highly selective for the COX-2 isoform in human cells [33]. Furthermore, at the dose used, NS-398 had no effect on proliferation or apoptosis of Bic-1 cells, which express COX-1 but not COX-2 [21]. These findings suggest that the suppression of FGF-2 expression and associated antiproliferative and proapoptotic effects of NS-398 are a result of selective inhibition of COX-2 activity, and not a nonspecific inhibition of COX-1. COX-2 inhibitors have previously been shown to inhibit proliferation and induce apoptosis in COX-2 expressing esophageal cell lines, including Seg-1 Molecular Carcinogenesis DOI 10.1002/mc

[21–24]. However, the role of FGF-2 in this process has not previously been reported. COX-2 activity and prostaglandin E2 (PGE2) production have been linked to stimulation of cellular proliferation [35–37], angiogenesis [38,39], and to suppression of apoptosis [40–42]. NS-398 suppresses PGE2 levels and inhibits proliferation of cultured esophageal epithelial cells derived from Barrett esophagus, and exogenous PGE2 can alleviate the growth inhibition [24]. The effects of PGE2 are mediated by signaling via a family of G protein coupled PGE receptors, designated EP1-EP4, which activate or inhibit a number of signaling pathways including PKA and PKC mediated pathways [43]. The FGF-2 gene promoter contains a consensus AP-1 response element [44], and transcriptional activation and expression of AP-1 proteins via one of several MAP kinase cascades stimulates FGF-2 gene transcription [45]. The FGF-2 gene is responsive to both PKC [46] and PKA [47] signaling pathways. In human esophageal adenocarcinoma cells derived from Barrett esophagus, EP1 and EP4 receptor antagonists (but not EP2 antagonists) inhibited proliferation and migration [48]. The effect on FGF2 expression was not examined in that study. However, PGE2 has been reported to stimulate the expression of FGF-2 in rat Muller cells via a PKCdependent pathway implicating the EP1 receptor in this response [49]. Inhibition of endogenous FGF-2 activity in Seg-1 cells with a neutralizing antibody also resulted in a marked increase in apoptosis, suggesting that FGF-2 plays an autocrine or paracrine role in promoting adenocarcinoma cell survival. This observation is consistent with our recent report [9] that overexpression of FGF-2 is associated with increased risk for tumor recurrence and reduced survival in patients with esophageal adenocarcinoma. We next evaluated COX-2 (mRNA and protein) expression in a well-defined series of surgically resected EADC tissues, previously characterized for FGF-2 expression [9]. In keeping with our initial report, overexpression of FGF-2 mRNA was also a predictor of survival (Figure 6C), although this did not reach formal statistical significance in this subset of patients. Consistent with several recent reports evaluating tissue distribution and expression of COX-2 in EADC [14,50–56] we found that COX-2 mRNA was overexpressed in 55% of tumors, although no statistical associations were seen between COX-2 mRNA and protein expression, levels of FGF-2 expression, or with any clinicopathologic parameter or outcome. The discordance between the mRNA and immunoreactive protein levels may reflect the non-quantitative nature of the RT-PCR and immunostaining analyses, or the heterogeneity of the tissue samples. However, even accounting for differences in immunohistochemical technique and interpretation of tissue sections, a

COX-2 AND FGF-2 IN ESOPHAGEAL ADENOCARCINOMA

limited number of studies have reported strong expression of COX-2 protein (ranging from 23% to 75% of tumors) to be a predictor of poor survival after surgical resection of EADC [14,52–54]. Here, as seen in Figure 6D, a trend towards reduced postoperative survival for patients with strong expression of COX-2 protein (46% in this study) was also found, although this did not reach formal statistical significance, again most likely reflecting limited patient numbers. With increasing use of multimodality therapy to treat EADC, the recent report that patients with persistent high COX-2 protein (but not COX-2 mRNA) expression after induction (preoperative) chemoradiotherapy exhibited only minor histopathologic response and poor survival [14], suggests that COX-2 protein expression, categorized by stringent immunohistochemical criteria, may be a clinically relevant molecular marker in the future management of human EADC. ACKNOWLEDGMENTS This work was supported by a grant to PRM from the Canadian Institutes for Health Research and to JB from NSERC. MB-N was supported by a fellowship from the Nova Scotia Health Research Foundation. The authors thank Dr. David Beer for providing the Seg-1 and Bic-1 cell lines, and Ms. Heather Sams for assistance with immunohistochemistry. GAP and AGC were supported by Clinical Research Scholarships from the Faculty of Medicine, Dalhousie University.

REFERENCES 1. Blot WJ, McLaughlin JK. The changing epidemiology of esophageal cancer. Semin Oncol 1999;26:2–8. 2. Botterweck AA, Schouten LJ, Volovics A, Dorant E, van Den Brandt PA. Trends in incidence of adenocarcinoma of the oesophagus and gastric cardia in ten European countries. Int J Epidemiol 2000;29:645–654. 3. Spechler SJ. Clinical practice. Barrett’s Esophagus. N Engl J Med 2002;346:836–842. 4. McManus DT, Olaru A, Meltzer SJ. Biomarkers of esophageal adenocarcinoma and Barrett’s esophagus. Cancer Res 2004;64:1561–1569. 5. Wild CP, Hardie LJ. Reflux, Barrett’s oesophagus and adenocarcinoma: Burning questions. Nat Rev 2003;3:676– 684. 6. Powers CJ, McLesky SW, Wellstein A. Fibroblast growth factors, their receptors and signaling. Endocr Relat Cancer 2000;7:165–197. 7. Iida S, Katoh O, Tokunaga A, Terada M. Expression of fibroblast growth factor gene family and its receptor gene family in the human upper gastrointestinal tract. Biochem Biophys Res Commun 1994;199:1113–1119. 8. Lord RV, Park JM, Wickramasinghe K, et al. Vascular endothelial growth factor and basic fibroblast growth factor expression in esophageal adenocarcinoma and Barrett esophagus. J Thorac Cardiovasc Surg 2003;125:246–253. 9. Barclay C, Li AW, Geldenhuys L, et al. Basic fibroblast growth factor (FGF-2) overexpression is a risk factor for esophageal cancer recurrence and reduced survival, which is ameliorated by coexpression of the FGF-2 antisense gene. Clin Cancer Res 2005;11:7683–7691. Molecular Carcinogenesis DOI 10.1002/mc

979

10. Jones DA, Carlton DP, McIntyre TM, Zimmerman GA, Prescott SM. Molecular cloning of human prostaglandin endoperoxide synthase type II and demonstration of expression in response to cytokines. J Biol Chem 1993;268:9049– 9054. 11. Zimmermann KC, Sarbia M, Weber AA, Borchard F, Gabbert HE, Schror K. Cyclooxygenase-2 expression in human esophageal carcinoma. Cancer Res 1999;59:198–204. 12. Yu HP, Xu SQ, Liu L, et al. Cyclooxygenase-2 expression in squamous dysplasia and squamous cell carcinoma of the esophagus. Cancer Lett 2003;198:193–201. 13. Jiang JG, Tang JB, Chen CL, et al. Expression of cyclooxygenase-2 in human esophageal squamous cell carcinomas. World J Gastroenterol 2004;10:2168–2173. 14. Xi H, Baldus SE, Warnecke-Eberz U, et al. High cyclooxygenase-2 expression following neoadjuvant radiochemotherapy is associated with minor histopathologic response and poor prognosis in esophageal cancer. Clin Cancer Res 2005;11: 8341–8347. 15. Abdalla EK, Pisters PW. Staging and preoperative evaluation of upper gastrointestinal malignancies. Semin Oncol 2004; 31:513–529. 16. Abdalla SI, Sanderson IR, Fitzgerald RC. Effect of inflammation on cyclooxygenase (COX)-2 expression in benign and malignant oesophageal cells. Carcinogenesis 2005;26: 1627–1633. 17. Li Y, Wo JM, Ray MB, et al. Cyclooxygenase-2 and epithelial growth factor receptor up-regulation during progression of Barrett’s esophagus to adenocarcinoma. World J Gastroenterol 2006;12:928–934. 18. Oyama K, Fujimura T, Ninomiya I, et al. A COX-2 inhibitor prevents the esophageal inflammation-metaplasiaadenocarcinoma sequence in rats. Carcinogenesis 2005; 26:565– 570. 19. Wang D, Dubois RN. Prostaglandins and cancer. Gut 2006; 55:115–122. 20. Piazuelo E, Jimenez P, Lanas A. COX-2 inhibition in esophagitis, Barrett’s esophagus and esophageal cancer. Curr Pharm Des 2003;9:2267–2280. 21. Souza RF, Shewmake K, Beer DG, Cryer B, Spechler SJ. Selective inhibition of cyclooxygenase-2 suppresses growth and induces apoptosis in human esophageal adenocarcinoma cells. Cancer Res 2000;60:5767–5772. 22. Vona-Davis L, Riggs DR, Jackson BJ, McFadden DW. Antiproliferative and apoptotic effects of rofecoxib on esophageal cancer in vitro(1). J Surg Res 2004;119:143– 148. 23. Yu HP, Shi LY, Lu WH, Su YH, Li YY, Xu SQ. Expression of cyclooxygenase-2 (COX-2) in human esophageal cancer and in vitro inhibition by a specific COX-2 inhibitor, NS-398. J Gastroenterol Hepatol 2004;19:638–642. 24. Buttar NS, Wang KK, Anderson MA, et al. The effect of selective cyclooxygenase-2 inhibition in Barrett’s esophagus epithelium: An in vitro study. J Natl Cancer Inst 2002;94: 422–429. 25. Sawaoka H, Tsuji S, Tsujii M, et al. Cyclooxygenase inhibitors suppress angiogenesis and reduce tumor growth in vivo. Lab Invest 1999;79:1469–1477. 26. Sanchez-Fidalgo S, Martin-Lacave I, Illanes M, Motilva V. Angiogenesis, cell proliferation and apoptosis in gastric ulcer healing. Effect of a selective cox-2 inhibitor. Eur J Pharmacol 2004;505:187–194. 27. Guo JS, Cho CH, Lam Liu ES, Choy HT, Wang JY, Leung Koo MW. Antiangiogenic effect of a highly selective cyclooxygenase-2 inhibitor on gastric ulcer healing in rats. Toxicol Appl Pharmacol 2002;183:41–45. 28. Masferrer JL, Leahy KM, Koki AT, et al. Antiangiogenic and antitumor activities of cyclooxygenase-2 inhibitors. Cancer Res 2000;60:1306–1311. 29. Yoshida S, Amano H, Hayashi I, et al. COX-2/VEGFdependent facilitation of tumor-associated angiogenesis and tumor growth in vivo. Lab Invest 2003;83:1385–1394.

980

BAGUMA-NIBASHEKA ET AL.

30. Li AW, Too CKL, Knee R, Wilkinson M, Murphy PR. FGF-2 antisense RNA encodes a nuclear protein with MutT-like antimutator activity. Mol Cell Endocrinol 1997;133:177–182. 31. Li AW, Too CKL, Murphy PR. The basic fibroblast growth factor (FGF-2) antisense RNA (GFG) is translated into a MutTrelated protein in vivo. Biochem Biophys Res Commun 1996; 223:19–23. 32. Baguma-Nibasheka M, Li AW, Osman MS, et al. Coexpression and regulation of the FGF-2 and FGF antisense genes in leukemic cells. Leuk Res 2005;29:423–433. 33. Kato M, Nishida S, Kitasato H, Sakata N, Kawai S. Cyclooxygenase-1 and cyclooxygenase-2 selectivity of non-steroidal anti-inflammatory drugs: Investigation using human peripheral monocytes. J Pharm Pharmacol 2001;53: 1679–1685. 34. Cryer B, Feldman M. Cyclooxygenase-1 and cyclooxygenase2 selectivity of widely used nonsteroidal anti-inflammatory drugs. Am J Med 1998;104:413–421. 35. Souza RF, Shewmake K, Pearson S, et al. Acid increases proliferation via ERK and p38 MAPK-mediated increases in cyclooxygenase-2 in Barrett’s adenocarcinoma cells. Am J Physiol Gastrointest Liver Physiol 2004;287:G743–G748. 36. Kaur BS, Khamnehei N, Iravani M, Namburu SS, Lin O, Triadafilopoulos G. Rofecoxib inhibits cyclooxygenase 2 expression and activity and reduces cell proliferation in Barrett’s esophagus. Gastroenterology 2002;123:60–67. 37. Kase S, Osaki M, Honjo S, et al. A selective cyclooxygenase-2 inhibitor, NS398, inhibits cell growth and induces cell cycle arrest in the G2/M phase in human esophageal squamous cell carcinoma cells. J Exp Clin Cancer Res 2004;23:301–307. 38. Gallo O, Franchi A, Magnelli L, et al. Cyclooxygenase-2 pathway correlates with VEGF expression in head and neck cancer. Implications for tumor angiogenesis and metastasis. Neoplasia 2001;3:53–61. 39. Wilson KT. Angiogenic markers, neovascularization and malignant deformation of Barrett’s esophagus. Dis Esophagus 2002;15:16–21. 40. Sheng H, Shao J, Morrow JD, Beauchamp RD, DuBois RN. Modulation of apoptosis and Bcl-2 expression by prostaglandin E2 in human colon cancer cells. Cancer Res 1998; 58:362–366. 41. Sheng GG, Shao J, Sheng H, et al. A selective cyclooxygenase 2 inhibitor suppresses the growth of H-ras-transformed rat intestinal epithelial cells. Gastroenterology 1997;113:1883– 1891. 42. Kase S, Osaki M, Honjo S, et al. Expression of cyclooxygenase-2 is correlated with high intratumoral microvessel density and low apoptotic index in human esophageal squamous cell carcinomas. Virch Arch 2003;442:129–135. 43. Ashby B. Co-expression of prostaglandin receptors with opposite effects: A model for homeostatic control of autocrine and paracrine signaling. Biochem Pharmacol 1998;55:239–244.

Molecular Carcinogenesis DOI 10.1002/mc

View publication stats

44. Shibata F, Baird A, Florkiewicz RZ. Functional characterization of the human basic fibroblast growth factor gene promoter. Growth Factors 1991;4:277–287. 45. Faris M, Ensoli B, Kokot N, Nel AE. Inflammatory cytokines induce the expression of basic fibroblast growth factor (bFGF) isoforms required for the growth of Kaposi’s sarcoma and endothelial cells through the activation of AP-1 response elements in the bFGF promoter. AIDS 1998;12:19–27. 46. Murphy PR, Sato Y, Sato R, Friesen HG. Regulation of multiple basic FGF mRNA transcripts by protein kinase C activators. Mol Endocrinol 1988;2:1196–1201. 47. Moffett J, Kratz E, Myers J, Stachowiak EK, Florkiewicz RZ, Stachowiak MK. Transcriptional regulation of fibroblast growth factor-2 expression in human astrocytes: Implications for cell plasticity. Mol Biol Cell 1998;9:2269–2285. 48. Piazuelo E, Jimenez P, Strunk M, et al. Effects of selective PGE(2) receptor antagonists in esophageal adenocarcinoma cells derived from Barrett’s esophagus. Prostaglandins Other Lipid Mediat 2006;81:150–161. 49. Cheng T, Cao W, Wen R, Steinberg RH, LaVail MM. Prostaglandin E2 induces vascular endothelial growth factor and basic fibroblast growth factor mRNA expression in cultured rat Muller cells. Invest Ophthalmol Visual Sci 1998; 39:581–591. 50. Morris CD, Armstrong GR, Bigley G, Green H, Attwood SE. Cyclooxygenase-2 expression in the Barrett’s metaplasiadysplasia-adenocarcinoma sequence. Am J Gastroenterol 2001;96:990–996. 51. van der Woude CJ, Jansen PL, Tiebosch AT, et al. Expression of apoptosis-related proteins in Barrett’s metaplasia– dysplasia–carcinoma sequence: A switch to a more resistant phenotype. Hum Pathol 2002;33:686–692. 52. Lagorce C, Paraf F, Vidaud D, et al. Cyclooxygenase-2 is expressed frequently and early in Barrett’s oesophagus and associated adenocarcinoma. Histopathology 2003;42:457– 465. 53. Buskens CJ, Sivula A, van Rees BP, et al. Comparison of cyclooxygenase 2 expression in adenocarcinomas of the gastric cardia and distal oesophagus. Gut 2003;52:1678– 1683. 54. Heeren P, Plukker J, van Dullemen H, Nap R, Hollema H. Prognostic role of cyclooxygenase-2 expression in esophageal carcinoma. Cancer Lett 2005;225:283–289. 55. von Rahden BH, Stein HJ, Puhringer F, et al. Coexpression of cyclooxygenases (COX-1, COX-2) and vascular endothelial growth factors (VEGF-A, VEGF-C) in esophageal adenocarcinoma. Cancer Res 2005;65:5038–5044. 56. Brabender J, Marjoram P, Lord RV, et al. The molecular signature of normal squamous esophageal epithelium identifies the presence of a field effect and can discriminate between patients with Barrett’s esophagus and patients with Barrett’s-associated adenocarcinoma. Cancer Epidemiol Biomarkers Prev 2005;14:2113–2117.