(FGF-2) Overexpression Is a Risk Factor for Esophageal Cancer

Human Cancer Biology

Basic Fibroblast Growth Factor (FGF-2) Overexpression Is a Risk Factor for Esophageal Cancer Recurrence and Reduced Survival, which Is Ameliorated by Coexpression oftheFGF-2AntisenseGene Christie Barclay,1Audrey W. Li,1Laurette Geldenhuys,2 Mark Baguma-Nibasheka,1Geoffrey A. Porter,3 Paul J.Veugelers,4 Paul R. Murphy,1and Alan G. Casson2,3

Abstract Purpose:The basic fibroblast growth factor (FGF-2) gene is bidirectionally transcribed to gener- ate overlapping sense and antisense (FGF-AS) mRNAs. FGF-AS has been implicated in the post- transcriptional regulation ofFGF-2 expression. The aim ofthis study was to characterize FGF-2 and FGF-AS in esophageal cancer and to correlate their expression with clinicopathologicfindings and outcome. Experimental Design: Reverse transcription-PCR was used to study FGF-2 and FGF-AS mRNA expression (normalized to glyceraldehyde-3-phosphate dehydrogenase) in 48 esophageal cancers relative to matched histologically normal esophageal epithelia (internal control). We used Cox proportional hazards analysis to calculate hazard ratios for recurrence and survival of patients with underexpression relative to the overexpression ofFGF-2 and/or FGF-AS. Results: Overexpression ofFGF-2 mRNA, by comparison with tumors underexpressing FGF-2, was associated with significantly increased risk for tumor recurrence (hazard ratio, 3.80; 95% confidence interval, 1.64-8.76) and reduced overall survival (hazard ratio, 2.11;95% confidence interval, 1.0-4.58).When the effects of FGF-2 and FGF-AS were considered simultaneously, the association ofFGF-2 mRNA overexpression with recurrence and mortality was even more pronounced, whereas FGF-AS mRNA overexpression was associated with reduced risk for recurrence and improved survival. Conclusions: Overexpression ofFGF-2 mRNA is associated with tumor recurrence and reduced survival after surgical resection of esophageal cancer and that these risks are reduced in tumors coexpressing the FGF-AS mRNA. These data support the hypothesis that FGF-AS is a novel tumor suppressor that modulates the effect of FGF-2 expression and may have potential clinical application to the development ofnovel therapeutic strategies.

Cancer of the esophagus, one of the ten most frequent Although the incidence of esophageal squamous cell carcinoma malignancies worldwide, is a relatively uncommon tumor in has remained steady, incidence rates for adenocarcinomas of North America (1). However, over the past three decades, there the lower esophagus and esophagogastric junction have has been a marked change in the epidemiology of this disease. increased in excess of any other human solid tumor (2, 3). Furthermore, despite recent advances in multimodality therapy, the prognosis for patients with invasive esophageal malignancy Authors’ Affiliations: Departments of 1Physiology and Biophysics, 2Pathology, remains generally poor (4). Substantial progress in the and 3Surgery, Faculty ofMedicine, Dalhousie University, Halifax, Nova Scotia, treatment of esophageal malignancy requires a clearer under- Canada and 4Department ofPublic Health Sciences, University ofAlberta, standing of esophageal tumor biology and the incorporation Edmonton, Alberta, Canada of molecular biomarkers into clinically relevant treatment Received 4/7/05; revised 7/22/05; accepted 8/4/05. Grant support: Cancer Research Society, Inc., Canadian Institutes for Health strategies. Research, Cancer Research Training Program Award from the Dalhousie Cancer Basic fibroblast growth factor (FGF-2) is the prototypic Research Program (C. Barclay), Nova Scotia Health Research Foundation (M. member of a family of related genes encoding heparin-binding Baguma-Nibasheka), and Dalhousie University Faculty ofMedicine Clinical proteins with growth, antiapoptotic, and differentiation pro- Research scholarships (G.A. Porter and A.G. Casson). moting activity (5). FGF-2 is expressed in esophageal squamous The costs ofpublication ofthis article were defrayed in part by the payment ofpage charges. This article must therefore be hereby marked advertisement in accordance cell carcinoma cell lines (6) and is elevated in esophageal with 18 U.S.C. Section 1734 solely to indicate this fact. adenocarcinoma (7), suggesting an autocrine or paracrine role Requests for reprints: Alan G. Casson, Division ofThoracic Surgery, Queen in esophageal tumorigenesis. The FGF-2 gene maps to Elizabeth II Health Science Centre, Victoria Building 7S-013, 1278 Tower Road, chromosome 4q26. This region is a site of frequent gain or Halifax, Nova Scotia, Canada B3H 2Y9. Phone: 902-473-2281; Fax: 902-473- 4426; E-mail: [email protected]. loss in esophageal adenocarcinoma and its premalignant F 2005 American Association for Cancer Research. lesion, Barrett’s esophagus (8–10), suggesting a possible doi:10.1158/1078-0432.CCR-05-0771 structural basis for dysregulation of FGF-2 function.

www.aacrjournals.org 7683 Clin Cancer Res 2005;11(21) November 1, 2005 Downloaded from clincancerres.aacrjournals.org on October 1, 2021. © 2005 American Association for Cancer Research. Human Cancer Biology

Although the control of FGF-2 expression is poorly under- normal esophageal epithelium adjacent to the proximal resection stood, one intriguing possibility is regulation by an endoge- margin were snap frozen in liquid nitrogen and stored in our nous antisense RNA, FGF-AS (11). Like the majority of esophageal tumor bank at À80jC for subsequent molecular analysis. cis-antisense transcript pairs characterized to date, FGF-2 and All remaining esophageal tissues were processed according to standard FGF-AS are transcribed as 3V-to-3V overlaps; the sense and protocol by the Department of Pathology of the Faculty of Medicine of Dalhousie University (Halifax, Nova Scotia, Canada). Representative antisense RNAs are fully complementary over extensive regions sections were stained with H&E and examined by an independent V of their 3 untranslated regions and have been shown to form consultant histopathologist. Serial unstained formalin-fixed, paraffin- stable double-stranded RNA duplexes in vivo (12). The organi- embedded tissue sections were used for subsequent immunohisto- zation and sequence of the FGF-2 and FGF-AS genes has been chemical analysis. All tumors were staged according to the International highly conserved from amphibian to human, indicating a Union Against Cancer (UICC) classification based on pTNM subsets functional importance for this structural relationship (reviewed (24). Strict clinicopathologic criteria (25) were used to define primary in ref. 13). The inverse association of FGF-2 and FGF-AS mRNA esophageal adenocarcinomas (Siewert type I), thereby excluding expression observed in a variety of species supports the adenocarcinomas of the cardia (Siewert type II) or proximal (subcardia) hypothesis that FGF-2 may be regulated by interaction with the gastric tumors (Siewert type III). antisense RNA (14–18). We have shown recently that forced All participating patients gave full informed consent, and collection overexpression of the FGF-AS mRNA can effectively suppress and storage of resected esophageal tissues was in accordance with the 1998 Canadian Tri-Council Policy ‘‘Statement on Ethical Conduct for FGF-2 levels in stably transfected cells in vitro (11). In addition Research Involving Humans.’’ Approval to study banked esophageal to its putative role as a regulatory RNA, FGF-AS also encodes tissues was approved by the Health Sciences and Humanities Research the protein GFG/NUDT6, a member of the nudix family of Ethics Board at Dalhousie University (2002-539). nucleoside phosphohydrolases (19, 20). The nudix motif is Reverse transcription-PCR. Reverse transcription-PCR was used to characteristic of a diverse group of phosphohydrolase enzymes study FGF-2 and FGF-AS mRNA expression [normalized to glyceralde- active on nucleoside diphosphates linked to another moiety (x). hyde-3-phosphate dehydrogenase (GAPDH)] in tumors, relative to Several nudix motif proteins, including the human homologues matched histologically normal esophageal epithelia (internal control), of MutT and MutY, have been shown to play important roles in using techniques reported previously (11). RNA was extracted from prevention and repair of DNA transversion mutations (21). each esophageal tissue specimen using the TRIzol reagent (Life Although the physiologic function of GFG is unknown, we have Technologies, Burlington, Ontario, Canada) according to the manufacturer’s instructions. Briefly, the tissue was homogenized in 1 mL shown recently that it can partially complement MutT function À TRIzol per 0.1 g tissue. The homogenate was centrifuged and RNA in in mutT Escherichia coli (20). the supernatant was purified using sequential washing with chloroform, The primary objective of this study was to evaluate the isopropanol, and 75% ethanol. The isolated RNA was resuspended expression of FGF-2 and FGF-AS mRNAs and their cognate using RNase-free water and stored at À80jC. Total RNA yield was proteins in a well-characterized series of surgically resected determined spectrophotometrically. Reverse transcription of 5 Ag RNA human esophageal tissues and cell lines and to correlate these was done using Moloney murine leukemia virus reverse transcriptase findings with clinicopathologic features and outcome. (Promega, Madison, WI) according to the manufacturer’s instructions. Amplification was done using EnzyPlus 2000 polymerase (EnzyPol Ltd., London, Ontario, Canada) in a 25 AL reaction composed of Materials and Methods 10 pmol of each primer, 0.5 mmol/L deoxynucleotide triphosphates, 2.2 mmol/L MgCl2, and 1.5 units polymerase in 10Â buffer. Thermal Patients and tumors. The study population comprised a consecutive cycling conditions followed an initial denaturation at 94jC for 3 series of 48 patients with esophageal malignancy treated by a single minutes and 29 (GAPDH) or 39 (FGF-2/FGF-AS) cycles of denaturation surgeon (A.G.C.) between February 1997 and February 2003. All patients at 94jC for 30 seconds, annealing at 60jC for 30 seconds, and had a histologic diagnosis of esophageal carcinoma from biopsies elongation at 72jC for 55 seconds followed by a final extension at 72jC obtained at esophagogastroscopy. Preoperative staging comprised com- for 5 minutes. PCR primers for FGF-2 were 5V-GGCTTCTTCCTGCG- puted tomography of the chest and upper abdomen, and all patients were CATCCA-3V (forward) and 5V-GCTCTTAGCAGACATTGGAAGA-3V considered to have locally resectable tumors, with no clinical evidence of (reverse). PCR primers for FGF-AS were 5V-CTGCAGTACAGCAATG- distant visceral metastases. No patient received induction chemotherapy GCGA-3V (forward) and 5V-CCTACTTGATGTAAGCATATC-3V (reverse). or radiotherapy. Subtotal esophagectomy was done using a right PCR primers for GAPDH were 5V-GAGCTGAACGGGAAGCTCACTGGC- transthoracic (31 patients) or transhiatal (17 patients) approach. A 3V (forward) and 5V-CCATGAGGTCCACCACCCTGTTGC-3V (reverse). potentially curative resection was done, completely resecting all Amplification products (FGF-2, 356 bp; FGF-AS, 198 bp; GAPDH, 321 bp) macroscopic tumor, with the thoracic and abdominal esophagus and were resolved on 1.8% agarose gels and stained with ethidium bromide. the lesser curvature of the stomach, to achieve a minimum 5 cm distal Levels of FGF-2 and FGF-AS mRNA expression in tumors, relative to resection margin as reported previously (22). Regional lymph node matched histologically normal esophageal squamous epithelia, were stations were resected extensively (two-field) and mapped to document stratified initially as follows: underexpressed (ratio < 0.7), the same patterns of metastasis. Reconstruction of the upper gastrointestinal tract (ratio = 0.71-1.5), or overexpressed (ratio > 1.5). All PCR products were was achieved by transposing stomach through the posterior mediasti- obtained within the linear range of the reaction. All reverse num, with a cervical esophagogastrostomy (23). Postoperative follow-up transcription-PCR assays were done (at minimum) in duplicate on comprised three monthly office visits for the first 3 years and then every coded samples by a single graduate student (C.B.) without knowledge 6 months at the center where surgery was done. All reasonable attempts of clinical correlative and outcome data. were made to confirm tumor recurrence and/or metastasis cytologically Immunohistochemistry. A modified indirect immunoperoxidase assay or histologically using radiologic-guided fine needle aspiration or was used to study FGF-2 and GFG protein expression and distribution in endoscopic biopsy techniques. serial unstained formalin-fixed, paraffin-embedded tissue sections (4 Am Immediately following esophageal resection, and in collaboration thickness). Affinity-purified anti-FGF-2 polyclonal antibodies raised with a consultant pathologist, resected esophageal tissues were against a synthetic peptide corresponding to amino acids 40 to 63 of examined and representative sections of the primary tumor (comprising human FGF-2 were obtained from Oncogene Research Products >80% malignant cells, with minimal necrosis) and histologically (Cambridge MA). The polyclonal antibody against the COOH-terminal

ClinCancerRes2005;11(21)November1,2005 7684 www.aacrjournals.org Downloaded from clincancerres.aacrjournals.org on October 1, 2021. © 2005 American Association for Cancer Research. FGF-2 Expression in Esophageal Cancer

Ta b l e 1. Tumor stage and survival at 5 years

n (%) Disease-free survival (%) P Overall survival (%) P Tstage

pT1 3(6) 67 0.002 67 0.024

pT 2 4(8) 25 50

pT 3 37 (77) 24 33

pT 4 4(8) 0 0 N stage

pN0 16 (33) 45 0.0001 78 <0.0001

pN1 32 (67) 11 13 Overall (UICC) stage I2(4) 50 <0.0001 50 <0.0001 IIA/B 14 (29) 45 70 III 29 (60) 13 16 IV 3 (6) 0 0

NOTE: The number ofpatients with stage IIA disease was 13 (27%); the number ofpatients with stage IIB was 1 (2%).Three (6%) patients have M 1disease (stage IV) based on the finding of distant (nonregional) lymph node metastases.

domain of the FGF-AS protein (GFG) has been characterized previously (19). Briefly, after deparaffinization in xylene, sections were hydrated through a series of ethanol solutions of graded concentration. Endogenous peroxidase activity was quenched with 3.0% hydrogen peroxide, and sections were heated in 10 mmol/L citrate buffer at 95jC for 10 minutes (antigen retrieval). Sections were incubated overnight at 4jC with primary antibodies against FGF-2 at 1:50 dilution and GFG at 1:100 dilution in a high-humidity chamber. Subsequent steps were done using Universal LSAB plus and 3,3V-diaminobenzidine plus kits according to the manufacturer’s protocols (DAKO Corp., Carpinteria, CA). After counterstaining with hematoxylin and dehydration, coverslips were applied. 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 done by two investigators at a double-headed microscope, blinded to mRNA expression or associated clinicopathologic data. To overcome the issue of tissue hetero- geneity, a 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 previously (25). Overall accumulation of FGF-2 or GFG 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 cells with weak immunoreactivity). Tissues were considered positive with a composite score of 4 to 8. The subcellular distribution of each protein (cytoplasmic, nuclear, or both) was recorded. Cell lines and culture. Two human esophageal adenocarcinoma cell lines (Bic-1, Seg-1; a generous gift from David Beer, Ph.D., University of Michigan, Ann Arbor, MI) were grown as monolayers in DMEM with L-glutamine (Life Technologies/Invitrogen Corp., Carlsbad, CA) supple- mented with 10% fetal bovine serum and maintained in an humidified atmosphere with 5% CO2 at 37jC. Cells were grown to semiconfluence, and after extraction of RNA, reverse transcription-PCR was used to evaluate expression of FGF-2 and FGF-AS mRNA as described above. Immunofluorescent confocal microscopy was used to study FGF-2 and GFG protein expression as described previously (26) with minor modification. Briefly, cells at f50% confluence in culture flasks were trypsinized and reseeded into dishes of equal area containing Fig. 1. Kaplan-Meier survival curves for the entire study cohort (n =48)after microscope slides and allowed to attach for 18 hours before further surgical resection ofesophageal cancer. A, overall survival. B, disease-free survival.

www.aacrjournals.org 7685 Clin Cancer Res 2005;11(21) November 1, 2005 Downloaded from clincancerres.aacrjournals.org on October 1, 2021. © 2005 American Association for Cancer Research. Human Cancer Biology

years (median, 63 years). Tumor histology was squamous cell in 10 patients and primary esophageal adenocarcinoma (Siewert type I) in 38 patients. Fourteen (29%) tumors were well differentiated, 12 (25%) were moderately differentiated, and 22 (46%) were poorly differentiated. Tumor stage is summarized in Table 1, with associated 5-year disease-free and overall survival. As expected, significant correlations between tumor stage and outcome were seen. At the time of last follow- up, 16 (33%) patients remain disease free, with a median follow-up of 31.7 months. Overall survival and disease-free survival for all patients are presented in Fig. 1. Fibroblast growth factor mRNA expression in human esophageal tissues. Representative gels illustrating expression of FGF- 2 and FGF-AS mRNAs are shown in Fig. 2. Using criteria defined a priori to stratify levels of FGF-2 mRNA expression in Fig. 2. Left, representative gels illustrating FGF-2 mRNA overexpression and tumors (relative to matched histologically normal esophageal FGF-AS mRNA underexpression in the same esophageal tumor (T) relative to matched normal (N) esophageal squamous epithelium. Right, range ofFGF-2 and squamous epithelia), FGF-2 mRNA was found to be underex- FGF-AS mRNA expression in esophageal cancers (normalized to GAPDH) relative pressed in 19 (40%) tumors, overexpressed in 17 (35%) to matched normal epithelia. Underexpression was defined aprioriby a relative tumors, and unchanged in 12 (25%) tumors. From the scatter mRNA level of <0.7, the same was defined as between 0.71and 1.5, and overexpression was defined as >1.50. From this scatter graph, two groups oftumors graph (Fig. 2) illustrating the range of FGF-2 mRNA expression, were identified based on FGF-2 and FGF-AS mRNA expression: underexpressors we identified two groups of tumors: those underexpressing and tumors with the same/overexpression. Horizontal bars, means for FGF-2 mRNA (n = 19) and those with the same/overexpression underexpression and overexpression for FGF-2 and FGF-AS. of FGF-2 mRNA (n = 29) relative to matched normal epithelia. manipulation. The cells were washed twice with cold PBS, fixed with Antisense fibroblast growth factor mRNA expression in human À20jC acetone for 2 minutes, air-dried, and kept at À70jC until further esophageal tissues. Relative to matched normal esophageal processing. The slides were treated with ice-cold 2% paraformaldehyde squamous epithelia, FGF-AS mRNA was underexpressed in (with 9 mg/mL disodium hydrogen orthophosphate and 6 mg/mL 17 (35%) tumors, overexpressed in 17 (35%) tumors, and L-lysine; pH 7.4) for 15 minutes, permeabilized with 0.1% Triton X-100 unchanged in 14 (30%) tumors. As noted for FGF-2 (see in PBS for 15 minutes, and then blocked with 3% bovine serum above), based on the range of FGF-AS mRNA expression (scatter albumin in PBS for 60 minutes. Double staining was by sequential graph; Fig. 2), two groups of tumors were identified: those exposure to primary antibodies and their corresponding fluorescently underexpressing FGF-AS mRNA (n = 17) and those with the tagged secondary antibodies each for 1 hour at room temperature. All same/overexpression of FGF-AS mRNA (n = 31) relative to antibodies were diluted in 0.1% bovine serum albumin-PBS as follows: matched normal epithelia. anti-FGF-2, 10 Ag/mL; anti-GFG, 350 Ag/mL; anti-nucleolin, 1.0 Ag/mL; Alexa Fluor 488 (green) or 594 (red) anti-rabbit and anti-goat Fibroblast growth factor and antisense fibroblast growth factor mRNA and protein expression in human esophageal cell lines. fluorescent F(abV)2 fragment IgG conjugates, 40 Ag/mL. Control staining to eliminate antibody nonspecificity was done by application of Figure 3 illustrates the relative expression of FGF-2 and FGF-AS secondary antibodies without prior exposure of the cells to the prima- mRNAs in the Bic-1 and Seg-1 human adenocarcinoma cell ries or following incubation with the primary antibody’s preimmuniz- lines. Whereas FGF-2 mRNA was not expressed in Bic-1, the ing peptide. Slides were mounted in a drop of glycerol-PBS (Citifluor, FGF-2 transcript was expressed in Seg-1. Both cell lines Marivac, Halifax, Nova Scotia, Canada). Image analysis used the expressed FGF-AS mRNA. Immunofluorescent microscopy standard operating software on the Zeiss LSM 510 microscope. showed that FGF-2 protein was localized only to the cytoplasm, Data collection and statistical analysis. Clinicopathologic data and whereas GFG was localized to the cytoplasm and nucleus, outcomes were prospectively collected and recorded in a research specifically the nucleolus (Fig. 4). database. Follow-up was complete for all patients until February 2005. Differences in the frequency of FGF-2 and FGF-AS mRNA and protein Fibroblast growth factor and GFG protein expression in human expression, according to demographic and clinicopathologic factors esophageal tissues. Using immunohistochemistry, FGF-2 protein (age, gender, tumor differentiation, pT stage, pN stage, and UICC was not detected in any normal esophageal squamous epithelia stage), were tested with a m2 test, with a Fisher exact test used if a cell (Fig. 5A) but was overexpressed in 83% (40 of 48) of tumors, contained <5 patients. The prognostic importance of FGF-2 and FGF-AS where immunoreactivity was localized exclusively to the mRNA expression in tumors (underexpressed versus same versus cytoplasm (Fig. 5B and C). overexpressed, underexpressed/same versus overexpressed, and underexpressed versus same/overexpressed) for disease-free and overall 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 = 0.05 and all analyses were done using SPSS for Windows 9.0 (SPSS, Inc., Chicago, IL).

Results

Clinicopathologic features, staging, and outcome. The series Fig. 3. Representative gels illustrating relative FGF-2 and FGF-AS mRNA comprised 41 males and 7 females ranging in age from 38 to 81 expression in two human esophageal adenocarcinoma cell lines, Seg-1and Bic-1.

Clin Cancer Res 2005;11(21) November 1, 2005 7686 www.aacrjournals.org Downloaded from clincancerres.aacrjournals.org on October 1, 2021. © 2005 American Association for Cancer Research. FGF-2 Expression in Esophageal Cancer

Fig. 4. Confocal immunofluorescence microscopy illustrating subcellular distribution ofFGF-2 and GFG protein in the human adenocarcinoma cell line, Seg-1.Top, immunofluorescent microscopy showed that FGF-2 protein was localized only to the cytoplasm, whereas GFG was localized to the cytoplasm and nucleus; bottom, nuclear GFG immunostaining was colocalized with nucleolin.

In normal esophageal squamous epithelia, weak immunore- FGF-2 overexpression on risk for tumor recurrence was even activity for GFG protein was seen only in the cytoplasm of cells more pronounced (HR, 5.45; 95% CI, 1.99-14.91; P = 0.001). in the middle third layer but not in cells of the basal or parabasal In contrast, FGF-AS mRNA overexpression was associated with layers. However, occasional cell nuclear immunoreactivity was a trend toward reduced risk for tumor recurrence (HR, 0.53; seen in proliferating basal and parabasal cells (Fig. 5D). In con- 95% CI, 0.21-1.32; P = 0.170) and improved overall survival trast, GFG protein was detected in the cytoplasm of all tumors (HR, 0.81; 95% CI, 0.34-1.90; P = 0.620). and in 63% (30 of 48) of tumor cell nuclei (Fig. 5E and F). On multivariate analysis, HRs did not substantially alter Associations between fibroblast growth factor and antisense when adjusted for age, gender, tumor histology, tumor fibroblast growth factor mRNA expression, survival, and clinico- differentiation, pT stage, pN stage, and overall (UICC) stage pathologic findings. Associations between levels of expression and when FGF-2 and FGF-AS mRNA expression in tumors of FGF-2 and FGF-AS mRNA in esophageal cancer (relative to (relative to matched normal) was stratified as same/underex- matched normal epithelia) and disease-free survival (reflecting pressed versus overexpressed or as underexpressed versus same tumor recurrence) and overall survival (mortality) are summa- versus overexpressed. The relationship between FGF-2 and rized in Table 2. Patients with tumors overexpressing FGF-2 FGF-AS mRNA expression and selected clinicopathologic mRNA had a significantly increased risk for tumor recurrence features of esophageal cancers in this series is shown in [hazard ratio (HR), 3.80; 95% confidence interval (95% CI), Table 3. 1.64-8.76; P = 0.0009] and reduced overall survival (HR, 2.11; As seen in Fig. 6C, based on coexpression of FGF-2 and FGF- 95% CI, 1.0-4.58; P = 0.050). Representative survival curves are AS mRNA in tumors, four subgroups of patients with different shown in Fig. 5A, illustrating improved disease-free survival for outcomes were identified, with FGF-2 underexpression/FGF-AS patients with tumors underexpressing FGF-2 mRNA. overexpression representing the best disease-free survival When the effect of FGF-2 and FGF-AS mRNA expression in and with FGF-2 overexpression/FGF-AS underexpression repre- tumors was considered simultaneously (Table 2), the effect of senting the worst disease-free survival.

Fig. 5. Immunohistochemistry was used to study tissue and subcellular distribution ofFGF-2 and GFG protein in esophageal tissues (normal and tumor) using a polyclonal anti-FGF-2 antibody at 1:50 dilution and a polyclonal antibody against the COOH-terminal domain ofthe GFG protein at 1:100 dilution. A, normal esophageal epithelium negative for FGF-2 protein (Â100). B, an esophageal adenocarcinoma with cytoplasmic positivity for FGF-2 protein (Â200). C, high-power view ofthe same adenocarcinoma with cytoplasmic positivity for FGF-2 (Â400). D, normal esophageal epithelium with weak cytoplasmic positivity for GFG in cells of the middle third layer ofthe epithelium and occasional nuclear positivity in cells ofthe basal and parabasal layers (Â100). E, an esophageal adenocarcinoma with cytoplasmic and strong nuclear positivity for GFG (Â200). F, high-power view ofthe same adenocarcinoma with cytoplasmic and strong nuclear positivity for GFG (Â400).

www.aacrjournals.org 7687 Clin Cancer Res 2005;11(21) November 1, 2005 Downloaded from clincancerres.aacrjournals.org on October 1, 2021. © 2005 American Association for Cancer Research. Human Cancer Biology

Ta b l e 2 . Associations (age and gender adjusted) between FGF-2 and FGF-AS mRNA expression (individually and simultaneously) and tumor recurrence and mortality after surgical resection of esophageal cancer

n Tumor recurrence Mortality HR (95 % CI) P HR (95 % CI) P

FGF-2 expression Underexpressed 19 ö 0.0009 ö 0.05 Same/overexpressed 29 3.80 (1.64-8.76) 2.11(1.0-4.58) FGF-AS expression Underexpressed 17 ö 0.47 ö 0.65 Same/overexpressed 31 1.32 (0.62-2.79) 1.21 (0.57-2.58) Simultaneous analysis of FGF-2 and FGF-AS FGF-2 expression Underexpressed 19 ö 0.001 ö 0.06 Same/overexpressed 29 5.45 (1.99-14.91) 2.34 (0.96-5.66) FGF-AS expression Underexpressed 17 ö 0.17 ö 0.62 Same/overexpressed 31 0.53 (0.21-1.32) 0.81 (0.34-1.90)

NOTE: Levels ofFGF-2 and FGF-AS mRNA expression (normalized to GAPDH, relative to matched normal esophageal squamous epithelia) were stratified as follows: underexpressed (ratio < 0.7) versus the same (ratio = 0.71-1.5)/overexpressed (ratio > 1. 51).

Discussion each with matched histologically normal epithelia. This series is unique as no patient received preoperative chemotherapy or In this present study, we first characterized the expression of radiation therapy, which is now current practice in many North FGF-2 and its natural antisense mRNA (FGF-AS) in a well- American centers and which may potentially confound defined series of surgically resected human esophageal tumors, molecular studies. We have shown, for the first time, that

Ta b l e 3 . Relationship ofselected clinicopathologic variables to FGF-2 and FGF-AS mRNA expression

FGF-2 mRNA same/overexpressed* FGF-AS mRNA same/overexpressed* n (%) Pn(%) P Gender Male 24/41 (59) 0.52 26/41 (63) 0.68 Female 5/7 (71) 5/7 (71) Tumor histology Squamous cell 5/10 (50) 0.45 6/10 (60) 0.73 Adenocarcinoma 24/38 (63) 25/38 (66) Tumor differentiation Well differentiated 8/14 (57) 0.56 11/14 (25) 0.73 Moderate 6/12 (50) 8/12 (67) Poor 15/22 (68) 12/22 (55) Overall (UICC) stage I 1/2 (50) 0.42 1/2 (50) 0.64 IIA/B 6/14 (43) 9/14 (64) III 20/29 (69) 20/29 (69) IV 2/3 (67) 1/3 (33) Tumor recurrence No 5/16(31) 0.0009 9/16(56) 0.47 Yes 24/32 (75) 22/32 (67) Overall survival Alive 8/17 (47) 0.05 10/17 (59) 0.62 Dead 21/31 (68) 21/31 (68)

*Same/overexpressed relative to matched normal epithelia.

ClinCancerRes2005;11(21)November1,2005 7688 www.aacrjournals.org Downloaded from clincancerres.aacrjournals.org on October 1, 2021. © 2005 American Association for Cancer Research. FGF-2 Expression in Esophageal Cancer

overexpression of FGF-2 mRNA is a significant predictor of tumor recurrence and mortality in a subset of patients with esophageal cancer and that coexpression of FGF-AS mRNA ameliorates this risk (Fig. 6C; Table 2). We also show that FGF- 2 protein was overexpressed in the Seg-1 esophageal adenocarcinoma cell line (Fig. 3) and in the majority of primary tumors (83%) where it was localized exclusively to the cytoplasm (Fig. 5). Ectopic expression of FGF-2 has been shown previously to transform normal cells into a malignant phenotype (27), and inhibition of FGF-2 expression with antisense oligonucleotides can reverse the transformed phenotype (28). The endogenous antisense RNA (FGF-AS) has been implicated in the post- transcriptional regulation of FGF-2 expression (11), but its possible role in tumor progression has not been investigated previously. The sense and antisense mRNAs are fully complementary over a span of several hundred nucleotides at their 3V ends and have been shown to form double-stranded RNA duplexes in vivo.InXenopus oocytes, where this phenomenon was first identified (12), the duplex was believed to be a target for ADAR-directed RNA editing and degradation. However, this sense-antisense interaction may also contribute to the post- transcriptional regulation of FGF-2 by several other possible mechanisms, including nuclear retention of the mRNA (29, 30), inhibition of pre-mRNA splicing, polyadenylation, RNA transport, or translation initiation (31–34). The inverse association of FGF-2 and FGF-AS mRNA levels in avian, rat, and human tissues, tumors, and cell lines has tended to corroborate the hypothesis that FGF-AS negatively regulates FGF-2 expression (14, 17, 35–37). We have shown previously that transfection and overexpression of FGF-AS in rat glioma cells resulted in post-transcriptional suppression of FGF-2 and inhibition of cell proliferation (11). It is therefore possible that the expression of the FGF-AS mRNA under basal conditions is sufficient to suppress FGF-2 gene expression by transcriptional or post-transcriptional mechanisms. We also showed here that GFG, the protein product of the FGF-AS mRNA, is expressed in esophageal tumor cells (Figs. 4 and 5). GFG/NUDT6 is a highly conserved protein belonging to the MutT/nudix family of nucleotide phosphohydrolases. The nudix box motif is a signature sequence characteristic of a family of enzymes that catabolize potentially toxic compounds in the cell (38, 39). The founding member of this family, the prokaryotic MutT protein, is responsible for removing 8-oxo- dGTP from the nucleotide pool, thus preventing transversion mutations caused by misincorporation of 8-oxo-guanine residues into DNA (reviewed in ref. 40). At least 20 distinct Fig. 6. Kaplan-Meier disease-free survival curves for patients after surgical human nudix motif genes are present in the human genome, 11 resection ofesophageal cancer categorized by underexpression versus same/ overexpression ofFGF-2 and FGF-AS or when both were considered of which (NUDT1-NUDT11) have been characterized to date. simultaneously. A, improved disease-free survival was seen for patients whose NUDT2, NUDT3, NUDT4, NUDT9, NUDT10, and NUDT11 tumors underexpressed FGF-2 mRNA (FGF decreased) in comparison with patients whose tumors overexpressed FGF-2 mRNA (P =0.0009).B, no significant encode enzymes active on a range of dinucleotides and the association between FGF-AS mRNA expression and disease-free survival was seen. structurally unrelated diphosphoinositol polyphosphates C, survival curves showing the effect of FGF-2 and FGF-AS mRNA expression (41–46). Dcp2 is a nudix protein implicated in the control in tumors when analyzed simultaneously. Four subgroups ofpatients were identified with different outcomes: FGF-2 underexpressed and FGF-AS of mRNA decay pathways (47). The MutT-related nudix overexpressed (FGF-2À, FGF-AS+; -----), FGF-2 underexpressed and FGF-AS proteins MTH1/NUDT1 (the human MutT homologue) and underexpressed (FGF-2À,FGF-ASÀ; ______), FGF-2 overexpressed and FGF-AS NUDT5 degrade oxidized and potentially mutagenic purine overexpressed (FGF-2+, FGF-AS+; ...... ), and FGF-2 overexpressed and FGF-AS underexpressed (FGF-2+, FGF-ASÀ; -.-.-.-.-.-.-.).When the effects of nucleoside diphosphates and triphosphates to the corre- FGF-2 and FGF-AS were considered simultaneously (Table 2), the effect of FGF-2 sponding monophosphates. The recent demonstration that overexpression was even more pronounced, with an increased risk for tumor MTH-1 knockout mice have a 2- to 3-fold increase in the ecurrence (HR, 5.45; 95% CI, 1.99-14.91; P = 0.001), whereas overexpression of FGF-AS mRNA was associated with a trend toward reduced risk for tumor frequency of adenomas and carcinomas of the lung, liver, and recurrence (HR, 0.53; 95% CI, 0.21-1.32; P =0.170). stomach supports the hypothesis that nudix proteins also play

www.aacrjournals.org 7689 Clin Cancer Res 2005;11(21) November 1, 2005 Downloaded from clincancerres.aacrjournals.org on October 1, 2021. © 2005 American Association for Cancer Research. Human Cancer Biology an antimutator role in mammalian cells (48). Although the (50–53), with the conclusion that only modest survival physiologic substrate of GFG (NUDT6) is not yet known, we advantage has been achieved using this approach (54, 55). As have established that the nudix domain of GFG is enzymatically FGF-2 has been shown recently to modulate sensitivity of tumor active and has antimutator activity that can partially comple- cells to various cytotoxic agents (56, 57), we have evaluated ment MutT-deficient E. coli (20). In human leukemic cells, recently the effect of FGF-2 on chemosensitivity of human GFG/NUDT6 is rapidly translocated from cytoplasm to nucleus esophageal adenocarcinoma cell lines exposed to cisplatin, a in response to mitogenic stimulation or stress, suggesting a key chemotherapeutic agent widely used in current clinical practice. role in the regulation of cellular proliferation (26). We showed Our preliminary data indicate that exogenous FGF-2 confers here that, although GFG protein was localized to the nucleus of resistance to cisplatin in Bic-1, which normally does not express occasional cells of the proliferative basal/parabasal layers of FGF-2 (Fig. 2), whereas no effect was seen in Seg-1 (58). normal esophageal squamous epithelium, GFG protein was In summary, we characterized the expression of FGF-2 and its detected in 63% of esophageal tumor cell nuclei. Although we natural antisense mRNA in a well-defined series of surgically cannot rule out the possibility that the protective effect of FGF- resected human esophageal cancers and have shown that AS is mediated by actions of GFG, the lack of association overexpression of FGF-2 mRNA, by comparison with tumors between GFG protein expression and recurrence or survival underexpressing FGF-2, was associated with significantly suggests that the protective effect is likely to be the result of the increased risk for tumor recurrence and reduced survival. These known post-transcriptional effects of the antisense RNA on observations have potential clinical application to improve the FGF-2 expression (11). accuracy of tumor staging, for prognosis, and to develop a Despite the striking changes reported recently for the rational basis for use of chemotherapy in the treatment of epidemiology of esophageal cancer (1–3), esophageal cancer esophageal malignancy (58). When the effects of FGF-2 and is a relatively uncommon tumor in North America, with FGF-AS were considered simultaneously, the association of incidence rates generally of <10 per 100,000 population. Given FGF-2 mRNA overexpression with recurrence and mortality was the rarity of this malignancy, the 48 tumors studied is a even more pronounced, whereas FGF-AS mRNA overexpres- relatively large series, which is representative of the provincial sion, but not its cognate protein product, ameliorates this risk. population. Additional strengths of this study include that all These data support the hypothesis that FGF-AS is a novel tumor tumors were treated in a consistent manner by a single suppressor that modulates the effect of FGF-2 expression. Lack university-based surgeon and that no patient received preoper- of association between GFG protein expression and recurrence ative chemotherapy or radiation therapy, that all tumors were or survival suggests that the protective effect may be the result well staged pathologically with primary esophageal adenocar- of post-transcriptional effects of the antisense mRNA on FGF-2 cinomas defined according to strict clinicopathologic criteria expression. (22, 23, 25), and that follow-up and outcomes data were complete for all patients. Acknowledgments Because of high rates of distant metastatic failure (89% in our series; ref. 25) after potentially curative esophageal resection, We thank the following individuals for help with selected aspects of this study: there has been considerable recent interest in the use of systemic Dr. David Beer for providing the Bic-1and Seg-1cell lines, Dr. Zuoyu Zheng for performing the immunohistochemical studies, Dr. S. Jane Darnton for obtaining chemotherapy in combination with surgery or radiation therapy follow-up on patients in this series, and the late Prof. John Crocker and Drs. Dickran (49). However, conflicting results have been obtained from Malatjalian and Heidi Sapp for expert independent histopathologic review of a limited number of randomized controlled clinical trials tissue sections.

References 1. Enzinger PC, Mayer RJ. Esophageal cancer. N Engl J volvement ofsequences on the long arm ofchromo- 15. Savage MP, Fallon JF. FGF-2 mRNA and its anti- Med 2003;349:2241^52. some 4. Cancer Res 1996;56:4499 ^ 502. sense message are expressed in a developmentally 2. Blot WJ, McLaughlin JK.The changing epidemiology 9. Riegman PHJ, Vissers KJ, Alers JC, et al. Genomic specific manner in the chick limb bud and mesoneph- ofesophageal cancer. Semin Oncol 1999;26:2 ^ 8. alterations in malignant transformation of Barrett’s ros. Dev Dyn 1995;202:343 ^ 53. 3. Powell J, McConkey CC, Gillison EW, Spychal RT. esophagus. Cancer Res 2001;61:3164 ^ 70. 16. Knee R, Li AW, Murphy PR. Characterization and tis- Continuing rising trend in oesophageal adenocarcino- 10. Doak SH, Jenkins GJS, Parry EM, et al. Chromo- sue-specific expression of the rat basic fibroblast ma. Int J Cancer 2002;102:422 ^ 7. some 4 hyperploidy represents an early genetic aber- growth factor antisense mRNA and protein. Proc Natl 4. Farrow DC, Vaughan TL. Determinants ofsurvival ration in premalignant Barrett’s oesophagus. Gut Acad Sci U S A1997;94:4943^7. following the diagnosis of esophageal adenocarcino- 2003;52:623 ^ 8. 17. Gagnon ML, Moy GK, Klagsbrun M. Characteriza- ma (United States). Cancer Causes Control 1996;7: 11. Li AW, Murphy PR. Expression ofalternatively tion ofthe promoter forthe human antisensefibroblast 322^ 7. spliced FGF-2 antisense RNA transcripts in the central growth factor-2 gene; regulation by Ets in Jurkat T 5. Powers CJ, McLesky SW, Wellstein A. Fibroblast nervous system: regulation ofFGF-2 mRNA transla- cells. J Cell Biochem1999;72:492 ^ 506. growth factors, their receptors and signaling. Endocr tion. Mol Cell Endocrinol 2000;170:233 ^ 42. 18. MihalichA,ReinaM,MangioniS,etal.Differentba- Relat Cancer 2000;7:165 ^ 97. 12. Kimelman D, Kirschner MW. An antisense messen- sic fibroblast growth factor and fibroblast growth fac- 6. Iida S, Katoh O,Tokunaga A,Terada M. Expression of ger RNA directs the covalent modification of the tran- tor-antisense expression in eutopic endometrial fibroblast growth factor gene family and its receptor script encoding fibroblast growth factor in Xenopus stromal cells derived from women with and without gene family in the human upper gastrointestinal tract. oocytes. Cell 1989;59:687 ^ 96. endometriosis. J Clin Endocrinol Metab 2003;88: Biochem Biophys Res Commun 1994;199:1113^ 9. 13. Murphy PR, Knee R, Li A. Regulation ofgene ex- 2853 ^ 9. 7. Lord RVN, Park JM,Wickamasinghe K, et al.Vascular pression in the CNS by natural antisense RNAs. In: 19. Li AW, Too CKL, Murphy PR. The basic fibroblast endothelial growth factor and basic fibroblast growth Leslie R, editor. Antisense knockdown in the CNS. growth factor (FGF-2) antisense RNA (GFG) is trans- factor expression in esophageal adenocarcinoma and Oxford: Oxford University Press; 1999. p.180 ^ 94. lated into a MutT-related protein in vivo.BiochemBio- Barrett esophagus. J Thorac Cardiovasc Surg 2003; 14. Knee RS, Pitcher SE, Murphy PR. Basic fibroblast phys Res Commun 1996;223:19^ 23. 125:246^ 53. growth factor sense (FGF) and antisense (GFG) RNA 20. Li AW,Too CKL, Knee R,Wilkinson M, Murphy PR. 8. Hammonud ZT, Kaleem Z, CooperJD, Sundaresan S, transcripts are expressed in unfertilized human FGF-2 antisense RNA encodes a nuclear protein with Patterson GA, Goodfellow PJ. Allelotype analysis of oocytes and in differentiated adult tissues. Biochem MutT-like antitumor activity. Mol Cell Endocrinol esophageal adenocarcinomas: evidence for the in- Biophys Res Commun 1994;205:577 ^ 83. 1997;133:177^ 82.

ClinCancerRes2005;11(21)November1,2005 7690 www.aacrjournals.org Downloaded from clincancerres.aacrjournals.org on October 1, 2021. © 2005 American Association for Cancer Research. FGF-2 Expression in Esophageal Cancer

21. CaiJ-P,KakumaT,TsuzukiT, Sekiguchi M. cDNA and 34. Singer R. RNA zipcodes for cytoplasmic addresses. revealed by Nudix motifhomology. Nature 2001; genomic sequences for rat 8-oxo-dGTPase that pre- Curr Biol 1993;3:719^ 21. 411:595 ^ 9. vents occurrence ofspontaneous mutations due to ox- 35. Murphy PR, Knee RS. Identification and character- 47. Piccirillo C, Khanna R, Kiledjian M. Functional char- idation ofguanine nucleotides. Carcinogenesis 1995; ization ofan antisense RNA transcript (GFG) from the acterization ofthe mammalian mRNA decapping en- 16:2243 ^ 50. basic fibroblast growth factor gene in human and rat zyme hDcp2. RNA 2003;9:1138 ^ 47. 22. Casson AG, Darnton SJ, Subramanian S, Hiller L. cells. Mol Endocrinol 1994;8:825^9. 48.TsuzukiT,EgashiraA,IgarashiH,etal.Spontaneous What is the optimum distal resection margin for esoph- 36. Asa SL, Ramyar L, Murphy PR, Li AW, Ezzat S. tumorigenesis in mice defective in the MTH1 gene ageal carcinoma? AnnThorac Surg 2000;69:205^9. The endogenous fibroblast growth factor-2 anti- encoding 8-oxo-dGTPase. Proc Natl Acad Sci U S A 23. Casson AG, Porter GA,Veugelers PJ. Evolution and sense gene product regulates pituitary cell growth 2001;98:11456^61. critical appraisal ofanastomotic technique following and hormone production. Mol Endocrinol 2001;15: 49. van Lanschot JJB, Gonzalez DG, Richel DJ. resection ofesophageal adenocarcinoma. Dis Esopha- 589^99. Surgery, radiotherapy, and chemotherapy for esoph- gus 2002;15:296^302. 37. Li A, Seyoum G, Shiu R, Murphy PR. Expression of ageal carcinoma. Curr Opin Gastroenterol 2001;17: 24. Sobin LH, Wittekind C, editors. In: TNM classifica- the rat bFGF antisense RNA transcript is tissue-specif- 400^5. tion ofmalignant tumors. 6th ed. New-York: Wiley- ic and developmentally regulated. Mol Cell Endocrinol 50.Walsh TN, Noonan N, Hollywood D, KellyA, Keeling Liss; 2002. p. 60 ^4. 19 9 6;118 :113 ^ 23. N, Hennessy TPJ. A comparison ofmultimodal thera- 25. Casson AG, Evans SC, Gillis A, et al. Clinical implica- 38.Volk R, Koster M, Poting A, Hartmann L, KnochelW. py and surgery for esophageal adenocarcinoma. N tions ofp53 tumor suppressor gene mutation and pro- An antisense transcript from the Xenopus laevis bFGF Engl J Med 1996;335:462 ^ 7. tein expression in esophageal adenocarcinomas: gene coding for an evolutionarily conserved 24 kd pro- 51. Bosset J-F, Gignoux M,Triboulet J-P, et al. Chemo- results ofa ten-year prospective study. J Thorac tein. EMBO J1989;8:2983^8. radiotherapy followed by surgery compared with sur- Cardiovasc Surg 2003;125:1121 ^ 31. 39. Bessman MJ, Frick DN, O’Handley SF. The MutT gery alone in squamous cell cancer ofthe esophagus. 26. Baguma-Nibasheka M, Li A, Osman MS, et al. proteins or ‘‘Nudix’’ hydrolases, a family of versatile, N Engl J Med 1997;337:161 ^ 7. Coexpression and regulation ofthe FGF-2 and FGF widely distributed, ‘‘housecleaning’’ enzymes. J Biol 52. Kelsen DP, Ginsberg R, PajakTF, et al. Chemothera- antisense genes in leukemic cells. Leuk Res 2005;29: Chem1996;271:25059^62. py followed by surgery compared with surgery alone 423 ^ 33. 40. Michaels ML, Tchou J, Grollman AP, Miller JH. for localized esophageal cancer. N Engl J Med 1998; 27. Quarto N,Talarico D, SommerA, Florkiewicz R, Basi- A repair system for 8-oxo-7,8-dihydrodeoxyguanine. 339:1979 ^ 84. lico C, Rifkin DB. 1989 Transformation by basic fibro- Biochemistry 1992;31:10964 ^ 8. 53. Medical Research Council Oesophageal Cancer blast growth factor requires high levels of expression: 41. Safrany ST, Caffrey JJ,Yang X, et al. A novel context Working Party. Surgical resection with or without pre- comparison with transformation by hst/K-fgf. Onco- for the ‘‘MutT’’ module, a guardian of cell integrity, in a operative chemotherapy in oesophageal cancer: a ran- gene Res 1989;5:101 ^ 10. diphosphoinositol polyphosphate phosphohydrolase. domized controlled trial. Lancet 2002;359:1727 ^ 33. 28. Murphy PR, SatoY, Knee R. Phosphorothioate anti- EMBO J1998;17:6599 ^607. 54. Kaklamanos IG, Walker GR, Ferry K, Franceschi D, sense oligonucleotides against basic fibroblast growth 42. Hua LV, Green M, Warsh JJ, Li PP. Molecular clon- Livingstone AS. Neoadjuvant treatment for resectable factor inhibit anchorage-dependent and anchorage- ing ofa novel isoform ofdiphosphoinositol poly- cancer ofthe esophagus and gastroesophageal junc- independent growth ofa malignant glioblastoma cell phosphate phosphohydrolase: a potential target of tion: a meta-analysis ofrandomized clinical trials. Ann line. Mol Endocrinol 1992;6:877 ^ 84. lithium therapy. Neuropsychopharmacology 2001;24: Surg Oncol 2003;10:754 ^ 61. 29. Kumar M, Carmichael GG. Antisense RNA: function 640^51. 55. UrschelJD,Vasan H. A meta-analysis ofrandomized and fate ofduplex RNA in cells ofhigher eukaryotes. 43. Caffrey JJ, Safrany ST,Yang X, Shears SB. Discov- controlled trials that compared chemoradiation and Microbiol Mol Biol Rev 1998;62:1415 ^ 34. ery ofmolecular and catalytic diversity among human surgery to surgery alone for resectable esophageal 30. Kumar M, Carmichael GG. Nuclear antisense RNA diphosphoinositol-polyphosphate phosphohydro- cancer. AmJ Surg 2003;185:538 ^ 43. induces extensive adenosine modifications and nucle- lases. An expanding nudt family. J Biol Chem 2000; 56. Song S, Wientjes MG, Gan Y, Au JLS. Fibroblast ar retention oftarget transcripts. Proc Natl Acad Sci 275:12730 ^ 6. growth factors: an epigenetic mechanism of broad U S A1997;94:3542^7. 44. McLennan AG. The MutT motiffamily ofnucleotide spectrum resistance to anticancer drugs. Proc Natl 31. McCarthyJEG, Kollmus H. Cytoplasmic mRNA-pro- phosphohydrolases in man and human pathogens [re- Acad Sci U S A 2000;97:8658 ^ 63. tein interactions in eukaryotic gene expression. TIBS view]. Int J Mol Med 1999;4:79 ^ 89. 57. Coleman AB. Positive and negative regulation ofcel- 19 9 5;20 :191 ^ 7. 45.Yang H, Slupska MM,WeiYF, et al. Cloning and char- lular sensitivity to anti-cancer drugs by FGF-2. Drug 32. Sachs A,Wahle E. Poly(A) tail metabolism and func- acterization ofa new member ofthe Nudix hydrolases Resist Updat 2003;6:85 ^ 94. tion in eukaryotes. J Biol Chem 1993;268:22955 ^ 8. from human and mouse. J Biol Chem 2000;275: 58. Li AW,Watanabe C, Casson AG, Murphy PR. FGF-2 33. Siebel C, Kanaar R, Rio D. Regulation oftissue-spe- 8844^53. modulates sensitivity to anticancer drugs in esopha- cific P-element pre-mRNA splicing requires the RNA- 46. Perraud AL, Fleig A, Dunn CA, et al. ADP-ribose geal adenocarcinoma cell lines [abstract]. Proc Am binding protein PSI. Genes Dev 1994;8:1713^ 25. gating ofthe calcium-permeable LTRPC2 channel Assoc Cancer Res 2005;46:LB263.

www.aacrjournals.org 7691 Clin Cancer Res 2005;11(21) November 1, 2005 Downloaded from clincancerres.aacrjournals.org on October 1, 2021. © 2005 American Association for Cancer Research. 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

Christie Barclay, Audrey W. Li, Laurette Geldenhuys, et al.

Clin Cancer Res 2005;11:7683-7691.

Updated version Access the most recent version of this article at: http://clincancerres.aacrjournals.org/content/11/21/7683

Cited articles This article cites 54 articles, 16 of which you can access for free at: http://clincancerres.aacrjournals.org/content/11/21/7683.full#ref-list-1

Citing articles This article has been cited by 4 HighWire-hosted articles. Access the articles at: http://clincancerres.aacrjournals.org/content/11/21/7683.full#related-urls

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

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

Permissions To request permission to re-use all or part of this article, use this link http://clincancerres.aacrjournals.org/content/11/21/7683. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.