FOXL1, a Novel Candidate Tumor Suppressor, Inhibits Tumor Aggressiveness and Predicts Outcome in Human Pancreatic Cancer

Published OnlineFirst June 25, 2013; DOI: 10.1158/0008-5472.CAN-13-0362

Cancer Molecular and Cellular Pathobiology Research

Geng Zhang1, Peijun He1, Jochen Gaedcke3, B. Michael Ghadimi3, Thomas Ried2, Harris G. Yfantis4, Dong H. Lee4, Nader Hanna5, H. Richard Alexander5, and S. Perwez Hussain1

Abstract The forkhead box L1 (FOXL1) transcription factor regulates epithelial proliferation and development of gastrointestinal tract and has been implicated in gastrointestinal tumorigenesis in mouse models. However, the role of FOXL1 in pancreatic cancer development and progression remains to be elucidated. Here, we report that higher expression of FOXL1 is significantly associated with better clinical outcome in human pancreatic ductal adenocarcinoma (PDAC). A lower FOXL1 expression is correlated with metastasis and advanced pathologic stage of pancreatic cancer. Mechanistic analyses showed that overexpression of FOXL1 induces apoptosis and inhibits proliferation and invasion in pancreatic cancer cells, whereas silencing of FOXL1 by siRNA inhibits apoptosis and enhances tumor cell growth and invasion. Furthermore, FOXL1 overexpression significantly suppressed the growth of tumor xenografts in nude mice. FOXL1 promoted apoptosis partly through the induction of TNF-related apoptosis-inducing ligand (TRAIL) in pancreatic cancer cells. In addition, FOXL1 suppressed the transcription of zinc finger E-box–binding homeobox 1 (ZEB1), an activator of epithelial–mesenchymal transition, and the negative regulation of ZEB1 contributed to the inhibitory effect of FOXL1 on tumor cell invasion. Taken together, our findings suggest that FOXL1 expression is a candidate predictor of clinical outcome in patients with resected PDAC and it plays an inhibitory role in pancreatic tumor progression. Cancer Res; 73(17); 1–10. 2013 AACR.

Introduction highly conserved 100-amino acid DNA-binding domain (the Pancreatic cancer is the fourth leading cause of cancer- forkhead box) and comprises more than 100 members in fi related death in the United States (1). The median survival of all humans, classi ed as FOXA to FOXR on the basis of sequence pancreatic ductal adenocarcinoma (PDAC) cases is less than 6 similarity (2). Fox proteins are at the junction of multiple months, and only 6% of patients survive 5 years after diagnosis. signaling pathways and play critical roles in a variety of The dismal prognosis in pancreatic cancer is due to late physiologic and pathologic processes including cancer. For diagnosis and the lack of effective treatment. A better under- example, FOXOs can initiate apoptosis and promote cell-cycle fi standing of the molecular mechanism of this disease and arrest (3). In addition, FOXOs de ciency in genetic mice led to discovery of novel therapeutic targets are desperately needed the development of thymic lymphomas and hemangiomas, to improve outcomes in patients with PDAC. indicating that the FOXOs are tumor suppressors (4, 5). In Forkhead box L1 (FOXL1) proteins belong to the forkhead contrast to FOXOs, FOXM1 has been shown to have propro- FOXM1 box (Fox) family of transcription factors. Fox family shares a liferative function, and increased expression of gene was often found in various human cancers (6). In pancreatic cancer, overexpression of FOXM1 is associated with poor Authors' Affiliations: 1Pancreatic Cancer Unit, Laboratory of Human prognosis and pathologic stage of PDAC (7). Downregulation Carcinogenesis, Center for Cancer Research and 2Genetics Branch, of FOXM1 results in the inhibition of migration, invasion, and 3 National Cancer Institute, NIH, Bethesda, Maryland; Department of Gen- angiogenesis of PDAC (8). eral and Visceral Surgery, University Medicine, Gottingen,€ Germany; 4Pathology and Laboratory Medicine, Baltimore Veterans Affairs Medical FOXL1 has been implicated in the regulation of epithelial cell Center; and 5Division of Surgical Oncology, The Department of Surgery and proliferation in gastrointestinal tracts. Loss of Foxl1 led to a the Marlene and Stewart Greenebaum Cancer Center, University of Mary- land School of Medicine, Baltimore, Maryland marked increase in cellular proliferation of intestinal epithelia in mice, leading to the distortion in the tissue architecture of Note: Supplementary data for this article are available at Cancer Research Online (http://cancerres.aacrjournals.org/). the stomach and small intestine (9). The altered proliferation rate in Foxl1-null–mutant mice is correlated with an activated Corresponding Author: S. Perwez Hussain, Pancreatic Cancer Unit, b Laboratory of Human Carcinogenesis, National Cancer Institute, NIH, 37 Wnt/ -catenin pathway as showed by increased nuclear trans- Convent Drive, Building 37, Room 3044B, Bethesda, MD 20892. Phone: location of b-catenin (10). Foxl1 deficiency accelerates the 301-402-3431; Fax: 301-496-0497; E-mail: [email protected] initiation of gastrointestinal tumor and increases tumor load Min doi: 10.1158/0008-5472.CAN-13-0362 in Apc mice (11). FOXL1 is specifically expressed in low- 2013 American Association for Cancer Research. grade fibromyxoid sarcoma (LGFMS), as compared with other

www.aacrjournals.org OF1

Zhang et al.

morphologically similar tumor types including myxofibrosar- and diaminobenzene as the chromogen (Dako Envision Sys- coma, desmoid fibromatosis, extraskeletal myxoid chondro- tem). Immunostaining was evaluated blindly by two Board- sarcoma, and solitary fibrous tumors, suggesting its role in certified pathologists assigning the intensity and prevalence LGFMS (12). Thus far, the biologic function of FOXL1 in human score as described elsewhere (14). Briefly, the intensity was cancer remains to be elucidated. In the present study, we assigned a score of 0–3, representing negative, weak, moderate, investigated the role of FOXL1 in human pancreatic cancer. or strong expression, whereas, prevalence was assigned a score Our data showed that FOXL1 expression is associated with of 0–4 representing less than 10%, 10%–30%, 30%–50%, 50%– clinical outcomes, pathologic stages, and metastasis of pan- 80%, and more than 80% cells showing FOXL1 expression. The creatic cancer. Furthermore, mechanistic studies showed that overall quantitation of IHC score was then achieved by mul- FOXL1 inhibits cellular proliferation and invasion in pancre- tiplying the intensity and prevalence score as described else- atic cancer cells. Moreover, we investigated the potential where (15). IHC images were photographed under an Olympus transcriptional targets of FOXL1, which may contribute to the BX40 microscope. inhibitory effects of FOXL1 on the growth and invasion in human pancreatic cancer cells. Cell lines and culture conditions Human pancreatic carcinoma cell lines Panc1 (CRL-1469), Materials and Methods MIApaca2 (CRL-1420), and Capan-1 (HTB-79), were obtained from American Type Culture Collection. Cells were maintained Tissue collection and RNA isolation in GIBCO RPMI Media 1640 supplemented with GlutaMAX-I Pairs of primary pancreatic tumor and adjacent nontumor (Invitrogen), penicillin–streptomycin (50 IU/mL and 50 mg/mL, tissues came from 45 patients with PDAC at the Department of respectively), and 10% (v/v) fetal calf serum. Cells were incu- General and Visceral Surgery, University Medicine Gottingen€ bated at 37C in a humidified atmosphere with 10% CO . (Gottingen,€ Germany). Tissues were flash frozen immediately 2 Transfection of the plasmid was conducted using Lipofecta- after surgery. Demographic and clinical information for each mine LTX reagent according to the manufacturer's protocol case included age, sex, grade, clinical staging, resection margin (Invitrogen). Plasmid pCMV-CTRL was used for transfection as status, and survival times from diagnosis (Supplementary the negative control. Plasmid pCMV-FOXL1 was used for Table S1). Tumor histopathology was classified according to FOXL1 overexpression. ON-TARGETplus siRNAs targeting to the World Health Organization Classification of Tumor sys- FOXL1 or zinc finger E-box–binding homeobox 1 (ZEB1) and tem (13). Use of these clinical specimens was reviewed and DharmaFECT-4 (Dharmacon) were used for silencing FOXL1 permitted by the National Cancer Institute (NCI)-Office of the or ZEB1 expression in pancreatic cancer cells. Human Subject Research (OHSR, Exempt # 4678) at the NIH (Bethesda, MD). RNA from frozen tissue samples was extracted MTT assay using standard TRIzol (Invitrogen,) protocol. RNA quality was Cells were seeded in 96-well plates (3,000 cells/well) and confirmed with the Agilent 2100 Bioanalyzer (Agilent Tech- incubated for 2 to 10 days. Then, the MTT solution was added nologies). and incubated for 4 hours. After the MTT solution was aspi- rated, 100 mL dimethylsulfoxide was added to each well. The Quantitative reverse transcriptase PCR absorbance was measured at 570 and 650 nm. Total RNA was reverse transcribed using High-Capacity cDNA Reverse Transcription Kit (Applied Biosystems). Quan- Apoptosis assay titative reverse transcriptase PCR (qRT-PCR) reactions in 384- Pancreatic cancer cell lines were seeded and transfected well plates were carried out using TaqMan gene expression with FOXL1 overexpression vectors or siRNAs directly target- assays on an ABI Prism 7900HT Sequence Detection instru- ing FOXL1 mRNA. Caspase activity was measured by Apo-ONE ment (Applied Biosystems). Expression levels of ACTB and Homogeneous Caspase-3/7 Assay (Promega) and potency of glyceraldehyde-3-phosphate dehydrogenase were used as the caspase activation was calculated and compared with control endogenous controls. All assays were conducted at least in cells. triplicates. For quality control, any samples with a gene cycle value more than 36 were considered of poor quality and Quantification of TRAIL protein level by ELISA removed. If a tumor or nontumor sample failed quality control Cells were transfected with either control vector pCMV- from qRT-PCR that case was removed from the analysis. All the CTRL or pCMV-FOXL1 expression vectors, respectively. Twen- primers for qRT-PCR in the present study were purchased from ty-four hours after transfection, cells were sorted and ali- Applied Biosystems. quoted in sample tubes (5 105cells/sample). Cells were lysed (5 105cells/100 mL lysate buffer) and TNF-related apoptosis- Immunohistochemistry inducing ligand (TRAIL) production in cell lysates was mea- Five-micrometer thick paraffin sections of tumors and sured using Quantikine Human TRAIL/TNFSF10 ELISA Kit surrounding nontumor tissues from resected PDAC cases were (R&D Systems) according to the manufacturer's instructions. incubated with mouse monoclonal anti-FOXL1 antibody Optical density of each well was then determined using a (Abnova). Signals were amplified using biotinylated immuno- microplate reader set to 450 nm. TRAIL concentrations were globulin G (IgG), followed by horseradish peroxidase-conju- calculated using a standard curve and linear regression gated avidin–biotin complex (Vectastain ABC Kit, Vector Lab) analysis.

OF2 Cancer Res; 73(17) September 1, 2013 Cancer Research

FOXL1 Suppresses Pancreatic Cancer Progression

Subcutaneous xenografts in nude mice cells were incubated for 10 minutes on a shaking platform at All animal experiments and maintenance conformed to room temperature. After the addition of glycine at concentra- the guidelines of the Animal Care and Use Committee and of tions of up to 0.136 mol/L, cells were incubated for 5 minutes the American Association of Laboratory Animal Care. A total and then washed with cold PBS. Cell pellets were obtained, of 5 106 FOXL1-overexpressing cells and control Panc1 resuspended in SDS lysis buffer, and sonicated with a Bioruptor cells in a total volume of 100 mL of 1/1 (v/v) PBS/Matrigel (Cosmo Bio,) for 30 seconds at maximum setting 10 times at 1- (BD Biosciences) were injected subcutaneously into flanks of minute intervals. Immunoprecipitation was carried out over- 8–9-week old male athymic nu/nu mice (Harlan Laborato- night at 4C with rotation, using antibodies specific for GFP ries) with 5 mice per arm. One week after the injection of (Origene) or mouse control IgG. After incubation with protein tumor cells, subcutaneous tumor volumes (V) were mea- A, the beads were collected, washed, and eluted per manu- sured weekly with digital calipers (Fisher Scientific) and facturer's instructions. Cross-links were reversed by incubation calculated using the formula V ¼ 1/2(ab2), where a is the for 6 hours at 65C and DNA was purified from the supernatant biggest and b is the smallest orthogonal tumor diameter. by phenol/chloroform extraction and ethanol precipitation. After 6 weeks, all mice were sacrificed and xenografts were Precipitated DNA was analyzed by qRT-PCR. The PCR primer resected and measured. Differences in tumor growth were pairs spanned upstream of promoter region of the human compared by the two-way repeated measures ANOVA and ZEB1 gene (GenBank accession number NM_001128128): (1) comparisons of tumor weight between the groups were forward, 50- TTCGAAGGGCTGGCATGGGT-30; reverse, 50- made using the two-tailed Student t test. P < 0.05 is con- CACCGTGGAACAAAGGAGGGCA-30; (2) forward, 50- GCAGG- sidered significant. GAGGTACAGGCCAGA-30; reverse, 50- TTGACCCCACCCCAC CAACA-30; (3) forward, 50-TGACCGCGTCCCTACGGTTT-30; Cell invasion assay reverse, 50- ACGGCCGGA ACCTTGTTGCT-30; (4) forward, Pancreatic cancer cell invasion assay was conducted in 24- 50-TTGTGCTGCTGTG CCAAGGGAA-30; reverse, 50-AAAGGC- well Biocoat Matrigel invasion chambers (8 mm; Becton Dick- GACTGTGCAACCACCA-30. The PCR primer pairs spanned inson) according to the manufacturer's protocol. Briefly, cells upstream of promoter region of the human TNFSF10 gene were transfected with either control vector pCMV-GFP or (TRAIL, GenBank accession number NM_003810.2): forward, pCMV-FOXL1-GFP expression vectors, respectively. Twenty- 50- TGCATGGATCCTGA GGGCAAGG -30; reverse, 50-TTGAA- four hours after transfection, the cells were harvested and CCTGCAACTGTCCCTCCC-30. qRT-PCR conditions were used: plated in the top chamber (5 104/well). The bottom chamber 35 cycles of 94C for 45 seconds, 60C for 45 seconds, and 72C contained 10% FBS as a chemoattractant. Forty-eight hours for 45 seconds. after incubation, the noninvasive cells were removed with a cotton swab. The invasive cells migrate through the membrane Statistical analysis and stick to the lower surface of the membrane. GFP-positive Expression graphs and Student t test were used to analyze fl fi cells were counted under a uorescence microscope in ve differences in gene expression using GraphPad Prism 5.0 fi fi elds ( 20 magni cation). Data are expressed as the percent- (GraphPad Software Inc). Kaplan–Meier analysis was con- age of invasion according to the manufacturer's manual: the ducted with GraphPad Prism 5.0. Cox proportional hazards number of the cells invading through the Matrigel membrane regression analysis was conducted using Stata 11 (StataCorp divided by the number of cells migrating through the control LP). Univariate Cox regression was conducted on genes and membrane. All assays were conducted in triplicate and repeat- clinical covariates to examine influence of each on patient ed three times. survival. Final multivariate models were based on stepwise addition and removal of clinical covariates found to be asso- Luciferase assay ciated with survival in univariate models (P < 0.05). For these ZEB1 TRAIL fi Human or promoter fragment was ampli ed models, resection margin status was dichotomized as positive from genomic DNA by PCR and cloned into pGL4Basic vector (R1) versus negative (R0); tumor-node-metastasis staging was fl m (Promega). Brie y, 1 g pCMV-FOXL1 (or pCMV-CTRL) and 1 dichotomized as I–II versus III–IV. All stepwise addition m fi fl gof re y-luciferase reporter plasmid (pGL4-ZEB or pGL4- models gave the same final models as stepwise removal Renilla TRAIL) together with 10 ng of a luciferase transfection models. All univariate and multivariate Cox regression models control (pRL-TK; Promega) were incubated overnight with were tested for proportional hazards assumptions based on fl subcon uent cell cultures in each well of a 24-well plate. The Schoenfeld residuals, and no model violated these assump- fi fl cells were then washed twice in PBS and harvested for re y/ tions. The statistical significance was defined as P < 0.05. All P Renilla luciferase assays using the Dual-Luciferase Reporter values reported were two-sided. Assay System (Promega).

Chromatin immunoprecipitation assays Results Chromatin immunoprecipitation (ChIP) assays were con- Higher FOXL1 is associated with better clinical outcome ducted according to the manufacturer's instructions (Qiagen). in pancreatic cancer Brieﬂy, Panc1 cells were transfected with pCMV-FOXL1-GFP. We conducted qRT-PCR for FOXL1 in 45 tumor samples of Forty-eight hours after transfection, formaldehyde was added PDAC and analyzed its association with disease outcome. We to cell culture media to a ﬁnal concentration of 1% and then the dichotomized high and low expression of FOXL1 as values

www.aacrjournals.org Cancer Res; 73(17) September 1, 2013 OF3

Zhang et al.

Univariate Cox regression analysis in PDAC cases showed A FOXL1 that a higher FOXL1 expression (HR, 0.44; 95% conﬁdence 100 P = 0.031 interval (CI), 0.20–0.95; P ¼ 0.036) and differentiation grade 80 (HR, 2.77; 95% CI, 1.12–6.87; P ¼ 0.027) were each associated 60 with prognosis but not the tumor stage or resection margin High n = 21 40 status (Table 1). Multivariate analyses revealed that higher FOXL1 expression was an independent predictor of better 20 Low n = 20 – P ¼

Percent survival % survival (HR, 0.41; 95% CI, 0.19 0.89; 0.024). In addition, 0 FOXL1 0204060 the multivariate model including and grade was sig- Months nificantly better at predicting prognosis in PDAC than grade B alone (P < 0.05, likelihood ratio test), indicating the prognostic significance of FOXL1 in resected PDAC. Immunohistochemical (IHC) analysis of tumors and adjacent nontumor tissues from PDAC cases showed predom- inant nuclear localization of FOXL1 in pancreatic ductal cells and acinar cells (Fig. 1B). PDAC cases with higher FOXL1 mRNA level determined by qRT-PCR also showed higher IHC staining score (Student t test, P < 0.05). FOXL1 expression was significantly higher in earlier stage (I þ II) as compared with the late stage tumors (III þ IV; Student t test, P < 0.01; Fig. 1C). Fisher exact test showed that a lowerexpressionofFOXL1issignificantly correlated with PDAC cases with lymph node and distant metastases (P < 0.01; Fig. 1D). These data indicated that FOXL1 may be involved in tumor progression and aggressiveness of pancreatic cancer. ** P < 0.01 CDHigh FOXL1 15 40 Low FOXL1 FOXL1 suppresses tumor growth of pancreatic cancer Fisher exact 30 The association of lower FOXL1 expression with 10 P < 0.01 increased cancer-specific mortality suggests its potential 20 inhibitory role in pancreatic tumor progression. To test this 5 IHC score 10 hypothesis, we first assessed the effect of FOXL1 on the Number of cases 0 0 proliferation and growth of pancreatic cancer cells. Panc1 MNM and MIApaca2 pancreatic cancer cells with a lower level of M: Metastasis endogenous FOXL1 expression (Supplementary Fig. S1) NM: Nonmetastasis were transfected with pCMV-FOXL1 expression construct. Nontumor (I/II) (III/IV) Early stage Late stage An elevated expression of FOXL1 was showed using qRT- PCR and Western blot analyses in cells transfected with Figure 1. Higher FOXL1 gene expression is associated with better clinical pCMV-FOXL1, as compared with control cells (Fig. 2A). outcome in PDAC. A, Kaplan–Meier analysis based on FOXL1 expression Overexpression of FOXL1 significantly inhibited cell prolif- levels in PDAC. FOXL1 gene expression values were determined by erationinbothPanc1andMIApaca2cells(P < 0.05; Fig. 2B). qRT-PCR and dichotomized into high and low groups using median value. fi fi The colony formation assay con rmed the inhibitory effect Survival pro les were compared using the log-rank test. B, representative fi images of IHC staining of FOXL1 in normal ductal cells (a), PanIN lesions (b), of FOXL1 on cell growth, showing a signi cant decrease in moderately differentiated tumors (c), and poorly differentiated tumors (d). colony number of FOXL1-overexpressing cells compared Original magnification, 200.Scale bar, 200 mm. C,FOXL1 immunostaining with control cells (P < 0.01, Fig. 2C). In contrast, knockdown is weaker in late stage (III/IV) as compared with early-stage (I/II) tumors of FOXL1 in pancreatic cancer cells using siRNA signi- (t test, P < 0.01). The intensity was assigned a score of 0–3, representing fi P < negative, weak, moderate, or strong expression, whereas prevalence was cantly increased cell growth ( 0.05, Supplementary assigned a score of 0–4, representing less than 10%, 10%–30%, 30%– Fig. S3C). 50%, 50%–80%, and more than 80% cells showed FOXL1 expression. The To further investigate the role of FOXL1 in tumor growth of overall quantitation of IHC score was then achieved by multiplying the pancreatic cancer in vivo, we extended our investigation by intensity and prevalence score as previously described. D, a lower subcutaneous implantation of FOXL1 overexpressing and expression of FOXL1 is associated with cases showing lymph node and distant metastases. Tumors were defined as high FOXL1-staining if their control Panc1 cells in nude mice. Tumor growth was signif- total IHC score was more than 6 and low staining if their score was less than icantly suppressed in cells overexpressing FOXL1 as com- or equal to 6. P value for Fisher exact test was indicated. pared with control cells (P < 0.05, Fig. 2D). Overexpression of FOXL1 led to a significant reduction in tumor volume (P < above and below the median, respectively, and found that a 0.05) and tumor weight (P < 0.01; Fig. 2D). These in vivo higher level of FOXL1 gene expression was associated with findings showed that FOXL1 suppresses the tumor growth of better survival (P ¼ 0.031, Kaplan–Meier log-rank test, Fig. 1A). pancreatic cancer cells.

OF4 Cancer Res; 73(17) September 1, 2013 Cancer Research

FOXL1 Suppresses Pancreatic Cancer Progression

Table 1. Cox regression analysis of FOXL1 expression with cancer-speciﬁc mortality in pancreatic cancer

Univariate analysis Multivariate analysisa

Variables (comparison/referent) HR (95% CI) P HR (95% CI) P FOXL1(high/low) 0.44 (0.20–0.95) 0.036 0.41 (0.19–0.89) 0.024 Grading (G3&4/1&2) 2.77 (1.12–6.87) 0.027 2.93 (1.20–7.15) 0.018 Tumor stage (III-IV/I-II) 2.51 (0.76–8.69) 0.130 Resection margin (R1/R0) 1.62 (0.75–3.51) 0.220

aMultivariate analysis used stepwise addition and removal of clinical covariates found to be associated with survival in univariate model and ﬁnal models include only those covariates that were signiﬁcantly associated with survival (P < 0.05).

FOXL1 increases TRAIL expression and promotes cells as compared with control cells (P < 0.01, Fig. 4A). In add- apoptosis ition, high level of FOXL1 expression also significantly impaired To determine the role of FOXL1 in regulating apoptosis in the invasiveness of both Panc1 and MIApaca2 in Matrigel inv- pancreatic cancer cells, we examined the caspase-3 activity in asion assays using 10% FBS as a chemoattractant (P < 0.01, Fig. FOXL1-overexpressing Panc1 and MIApaca2 cells. Overexpres- 4B). In contrast, knockdown of FOXL1 promoted pancreatic sion of FOXL1 led to a significant increase in caspase-3/7 activity cancer cell invasion (P < 0.05, Supplementary Fig. S3B). in Panc1 and MIApaca2 cells (P < 0.01, Fig. 3A), whereas silencing of FOXL1 by siRNA reduced apoptosis as determined by cas- FOXL1 binds to the promoter of ZEB1 and suppresses pase-3/7 activity (P < 0.01, Supplementary Fig. S3D). To elucidate ZEB1 transcription the underlying mechanism of the regulation of apoptosis by ZEB1 functions as an epithelial–mesenchymal transition FOXL1, we then analyzed expressions of a panel of apoptosis- (EMT) activator, promoting cancer cell invasion and aggres- related genes in response to the overexpression of FOXL1 using siveness (17). It has been reported that ZEB1 expression is qRT-PCR and found that FOXL1 overexpression induced proa- higher in Panc1 and MIApaca2, pancreatic cancer cell lines poptotic gene TRAIL expression by about 2.5-fold in pancreatic with mesenchymal features, as compared with HPAF, Capan1, cancer cell lines as compared with controls (P < 0.01; Fig. 3B). and Capan2 cell lines with epithelial features (18). We observed Consistent with the gene expression, the protein level of TRAIL an inverse association between FOXL1 and ZEB1 expressions was also increased about 2-fold in FOXL1-overexpressing cells. in pancreatic cancer cell lines (Supplementary Fig. S1A). Quan- Next, we investigated the mechanism of TRAIL induction titative RT-PCR and Western blot analyses showed that FOXL1 following FOXL1 expression. ChIP-PCR assays showed the was higher in HPAF, Capan1, and Capan2 cell lines and lower in binding of FOXL1 protein to the TRAIL promoter (P < 0.01; Fig. Panc1 and MIApaca2 cells (Supplementary Fig. S1). We then 3C). The promoter region (1,523 to þ23, relative to the tested the hypothesis that a higher level of FOXL1 suppresses transcription start site) of TRAIL was then inserted into the ZEB1 expression. Quantitative RT-PCR and Western blot anal- pGL4 basic luciferase reporter vector (pGL4-TRAIL) as previ- yses showed that overexpression of FOXL1 significantly inhib- ously described (16). The luciferase activities of pGL4-TRAIL ited ZEB1 expression in Panc1 and MIApaca2 cells (Fig. 5A). In were significantly upregulated by 4.4-fold (P < 0.01) when it was contrast, knockdown of FOXL1 increases the expression of cotransfected with pCMV-FOXL1 compared with pCMV-CTRL ZEB1 (P < 0.05, Supplementary Fig. S1). To determine whether in Panc1 cells and 5.7-fold in MIApaca2 cells (P < 0.01, Fig. 3D). ZEB1 is involved in inhibitory effect of FOXL1 on cell invasion These findings indicate that FOXL1 induces the expression and in pancreatic cancer, pancreatic cancer cell lines were trans- promoter activity of TRAIL. fected with si-FOXL1 alone or together with si-ZEB1. The We further defined the role of TRAIL in FOXL1-induced stimulatory effect of FOXL1 knockdown on cell invasion was apoptosis using an antibody directed against TRAIL in FOXL1- ablated by simultaneous siRNA-mediated knockdown of ZEB1 overexpressing cells. After transfection, Panc1 and MIApaca2 (Supplementary Fig. S3B). These data indicated that the neg- cells were treated with anti-TRAIL antibody (30 mg/mL). Anti- ative regulation of ZEB1 may contribute to the inhibitory effect TRAIL antibodies significantly blocked FOXL1-induced apo- of FOXL1 on invasion ability of pancreatic cancer cells. ptosis (Fig. 3E). Taken together, these results indicate that Analysis of ZEB1 promoter to identify potential forkhead FOXL1 promotes apoptosis in pancreatic cancer, at least box–binding sites using TRANSFAC program revealed 4 partly, through the induction of TRAIL expression. regions containing multiple potential FOXL1-binding sites at the upstream of transcription start site of ZEB1 (Fig. 5B). To FOXL1 inhibits migration and invasion ability of PDAC further investigate the mechanism of FOXL1-mediated regu- cells lation of ZEB1, ChIP-PCR assays were conducted to determine To investigate the effect of FOXL1 on cancer cell migration whether a physical interaction exists between FOXL1 protein and invasion, the monolayer scratch healing and Matrigel and the promoter region of ZEB1. We pulled down FOXL1- invasion assays were conducted. Quantitative analyses showed targeted DNA fragments and probed for 4 fragments repre- a significant reduction in wound closure in FOXL1-transfected senting different ZEB1 promoter regions by qRT-PCR (Fig. 5C).

www.aacrjournals.org Cancer Res; 73(17) September 1, 2013 OF5

Zhang et al.

Figure 2. FOXL1 inhibits pancreatic tumor growth. A, increased expression of FOXL1 in transfected pancreatic cancer cells were showed by qRT-PCR and Western blot analysis. B, there were signiﬁcant decreases in cell growth of FOXL1-overexpressing cells as compared with control cells. C, FOXL1-overexpressing cells exhibited reduced colony formation as compared with control cells. Data are presented as means SD from 3 independent experiments. , t test P < 0.05; , t test P < 0.01. D, subcutaneous implantation of FOXL1- overexpressing Panc1 cells in nude mice showed a signiﬁcant decrease in tumor growth as compared with control cells.

The strongest association was detected around region-4 within FOXL1 or pCMV control vector. As shown in Fig. 5D, the the ZEB1 promoter (P < 0.01). We next tested whether this luciferase activities of pGL4-ZEB1 were signiﬁcantly decreased interaction affects ZEB1 transcription. The promoter region of when it was cotransfected with pCMV-FOXL1 compared with ZEB1 harboring region-4 was inserted into the pGL4 basic pCMV-CTRL (P < 0.01, Fig. 5D). These ﬁndings showed that reporter vector (pGL4-ZEB1). Luciferase activity was deter- FOXL1 bound to the promoter region of ZEB1 and suppressed mined following cotransfection of pGL4-ZEB1 with pCMV- transcription and promoter activity of ZEB1.

OF6 Cancer Res; 73(17) September 1, 2013 Cancer Research

FOXL1 Suppresses Pancreatic Cancer Progression

Figure 3. FOXL1 induces TRAIL expression and promotes apoptosis. A, overexpression of FOXL1 led to a signiﬁcant increase in caspase-3/7 activity in Panc1 and MIApaca2 cells (P < 0.01). Relative caspase-3/7 activity represents the effect of FOXL1 overexpression on apoptosis compared with control cells at 24 hours and 48 hours posttransfection. B, FOXL1 overexpression upregulated TRAIL expression. Real-time PCR was conducted to determine TRAIL mRNA levels. TRAIL protein level in cell lysates was measured using ELISA kit (R&D Systems) according to the manufacturer's instructions. C, ChIP-PCR assay showed the binding of FOXL1 protein to the TRAIL promoter. ChIP assay was conducted in Panc1 and MIApaca cells, followed by qRT-PCR using a primer pair for TRAIL promoter. Lysates were immunoprecipitated with control mouse IgG and anti-GFP antibody for GFP-tagged FOXL1. RT-PCR values were normalized to the total amount of DNA added to the reaction (input). Data are presented as percent input. D, Luciferase reporter assays showed the stimulatory effect of FOXL1 on TRAIL transcription in Panc1 and MIApaca2 cells. The pGL4-TRAIL vector was cotransfected with FOXL1 expression vector pCMV-FOXL1 or empty vector pCMV-CTRL. Renilla luciferase transfection (pRL-TK) was also included in each assay for normalization in Dual-Luciferase Reporter Assay System (Promega). Each transfection was conducted in triplicates. , t test P < 0.01. E, anti-TRAIL antibodies blocked FOXL1-induced apoptosis. Data represent means SD from 3 independent experiments. , t test P < 0.01.

Discussion dataset showed a higher gene expression of FOXL1 in micro- FOXL1 was previously described as a critical transcriptional dissected early PanIN lesions as compared with microdis- factor that regulates cell proliferation and development of sected normal duct epithelium (27). In our study, IHC analysis, epithelium in gastrointestinal tracts in mice (9). We conducted which allows us to compare the FOXL1 expression in ductal Oncomine database analyses on publicly available microarray cells, showed that low FOXL1 expression is correlated with datasets and found that FOXL1 expression is consistently metastasis and advanced pathologic stage of pancreatic cancer decreased in multiple myeloma as compared with normal (Fig. 1). Moreover, our data showed for the first time that a tissues, with a cut-off P < 0.05 in microarray studies (19–21). higher FOXL1 expression is associated with better clinical out- One myeloma dataset also showed that FOXL1 is significantly come in resected PDAC cases. Functional studies revealed higher in long survival (alive over a year) as compared with that restoration of FOXL1 in Panc1 and MIApaca2 pancreatic short survival group (dead within a year; ref. 20). However, cancer cells significantly promoted apoptosis, inhibited cell differential expression of FOXL1 between pancreatic tumors proliferation, suppressed cell invasion, and inhibited Panc1 and nontumors in available microarray datasets from Onco- tumor growth in nude mice. Taken together, our findings mine are inconsistent (Supplementary Fig. S2). A lower FOXL1 support the inhibitory role of FOXL1 in pancreatic cancer. expression was found in pancreatic tumors as compared with Furthermore, we identified ZEB1 as a novel transcriptional nontumor pancreas tissues in two publicly available micro- target of FOXL1. ZEB1, encoded by the TCF8 gene, plays an array datasets (22, 23). Interestingly, Ishikawa and colleagues' important role in invasion and metastasis of human tumors data showed that FOXL1 level was significantly lower in pan- (17). Enforced expression of ZEB in epithelial cells results in a creatic juice from patients with pancreatic cancer than in rapid EMT associated with a breakdown of cell polarity, loss of healthy donors (24). In contrast, overexpression of FOXL1 in cell–cell adhesion, and induction of cell motility (28). Expres- tumors was found in Pei and colleagues' dataset (25) and no sion of ZEB1 promotes metastasis of tumor cells in a mouse significant difference exists in FOXL1 expression between xenograft model (29). Vice versa, knockdown of ZEB factors in tumor and normal tissues in Grutzmann and colleagues' data- cancer cells inhibit cell invasion (30). In this study, we showed set (26). Gene expression profile of Prasad and colleagues' that FOXL1 protein directly binds to ZEB1 promoter and

www.aacrjournals.org Cancer Res; 73(17) September 1, 2013 OF7

Zhang et al.

Figure 4. FOXL1 inhibits cell migration and invasion of pancreatic cancer cells. A, cell migration was assessed using scratch-healing assays. Conﬂuent monolayer of Panc1 and MIApaca2 cells were scratched and healing was monitored by taking photographs at the indicated time points. B, cell invasion was determined in Panc1 and MIApaca2 cells using Biocoat Matrigel invasion assay. The invaded cells were counted under a microscope. Data represent means SD from 3 independent experiments. , t test P < 0.01.

suppresses the transcription and promoter activity of ZEB1 TRAIL an attractive candidate for cancer therapy (41). TRAIL (Fig. 5). Our data indicate that inhibition of ZEB1 expression by has been reported to induce apoptosis in a wide spectrum of FOXL1 may contribute to the inhibitory effect of FOXL1 on cancer cell lines including pancreatic cancer cell lines, with no invasive ability of pancreatic cancer cells. or minimal toxicity on normal human cells (42–45). We found Apoptosis plays a critical role in tumorigenesis. Several that enforced expression of FOXL1 resulted in an increased members of Fox family have been linked to the regulation of TRAIL expression in pancreatic cancer cells, which eventually apoptosis in cancer (31). For example, FOXO proteins regulate led to the elevated caspase-3 activity in these cells (Fig. 3). cell survival by modulating the expression of death receptor There was no difference in TRAIL-R1 or R2 expression in ligands (such as FasL and TRAIL), which function in autocrine FOXL1-overexpressing cells as compared with control cells. and paracrine manners (16, 32). In addition to the death We also found approximately 1.5-fold increase in Fas expres- receptor ligands, FOXO proteins have been shown to be sion in these cells (Supplementary Fig. S4). However, several involved in the transactivation of the Bcl-2 family, which has studies have shown that PDAC is resistant to Fas/CD95- both pro- and antiapoptotic members and plays a critical role mediated apoptosis, even though both Fas receptor and Fas in regulating cell survival (33, 34). In our study, the induction of ligand are frequently expressed in PDAC cells (46, 47). Thus, apoptosis by FOXL1 seemed to be associated with upregulation increased Fas expressions have limited impact on the induc- of TRAIL expression. tion of apoptosis by FOXL1, at least in pancreatic cancer. TRAIL is a member of the TNF superfamily that is expressed In summary, our findings showed for the first time that in most human tissues (35, 36). The ligation of TRAIL with two FOXL1 plays an inhibitory role in pancreatic cancer. Over- receptors, called death receptor 4 (DR4, TRAIL-R1) and death expession of FOXL1 promoted apoptosis, suppressed cellular receptor 5 (DR5, TRAIL-R2; refs. 37–39), triggers apoptosis by growth and invasion in pancreatic cancer cells, at least partly recruiting the initiator caspase-8, which can directly activate through regulation of TRAIL and ZEB1 expression, and inhib- downstream effector caspases, including caspase-3, caspase-6, ited tumor growth in nude mice. These results are consistent and caspase-7 (40). The potent death-inducing ability makes with our finding that a higher FOXL1 expression is associated

OF8 Cancer Res; 73(17) September 1, 2013 Cancer Research

FOXL1 Suppresses Pancreatic Cancer Progression

Figure 5. FOXL1 suppresses expression and promoter activity of ZEB1. A, overexpression of FOXL1 results in a decrease in ZEB1 expression as analyzed by qRT- PCR and Western blotting. B, schematic representation of 50 upstream of transcriptional start site (TSS) of the ZEB1 gene. Predicted forkhead box–binding sites with good match were marked (Matrix similarity > 0.9). C, ChIP assay was conducted in Panc1 cells, followed by qRT-PCR using primer pairs spanning 4 regions in ZEB1 promoter. Lysates were immunoprecipitated with control mouse IgG and anti-GFP antibody for GFP-tagged FOXL1. Quantitative RT-PCR values were normalized to the total amount of DNA added to the reaction (input). Data are presented as percentage input. D, luciferase reporter assays showed the inhibitory effect of FOXL1 on ZEB1 transcription in Panc1 and MIApaca2 cells. The pGL4-ZEB1 vector was cotransfected with FOXL1 expression vector pCMV-FOXL1 or empty vector pCMV-ctrl. Renilla luciferase transfection (pRL-TK) was also included in each assay for normalization in Dual-Luciferase Reporter Assay System (Promega). Each transfection was conducted in triplicates. , t test P < 0.01.

with better clinical outcome in patients with PDAC. Therefore, Writing, review, and/or revision of the manuscript: S.P. Hussain, G. Zhang, J. Gaedcke, B.M. Ghadimi, D.H. Lee, H.R. Alexander we propose that FOXL1 may function as a potential tumor Administrative, technical, or material support (i.e., reporting or orga- suppressor and serve as a candidate predictor of outcomes in nizing data, constructing databases): P. He, T. Ried pancreatic cancer. Further studies are warranted to determine Study supervision: S.P. Hussain the molecular mechanism driving the loss of FOXL1 during pancreatic cancer progression and how the restoration of Acknowledgments The authors thank Drs. Stefan Ambs and Xin Wei Wang (National Cancer FOXL1 can be harnessed for therapeutic benefit. Institute) for helpful discussions. They also thank Ms. Elise Bowman for handling of clinical samples and maintenance of tissue database and Dr. Matthias Gaida Disclosure of Potential Conflicts of Interest for histopathologic evaluation. No potential conflicts of interest were disclosed. Grant Support Authors' Contributions This work is supported by Intramural Research Program of the National Conception and design: S.P. Hussain, G. Zhang Cancer Institute, Center for Cancer Research, NIH. The costs of publication of this article were defrayed in part by the payment of Development of methodology: G. Zhang, P. He advertisement Acquisition of data (provided animals, acquired and managed patients, page charges. This article must therefore be hereby marked in provided facilities, etc.): G. Zhang, J. Gaedcke, B.M. Ghadimi, H.G. Yfantis, accordance with 18 U.S.C. Section 1734 solely to indicate this fact. N. Hanna, H.R. Alexander Analysis and interpretation of data (e.g., statistical analysis, biostatistics, Received February 6, 2013; revised May 20, 2013; accepted May 30, 2013; computational analysis): G. Zhang published OnlineFirst June 25, 2013.

References 1. Siegel R, Naishadham D, Jemal A. Cancer statistics, 2012. CA Cancer 4. Paik JH, Kollipara R, Chu G, Ji H, Xiao Y, Ding Z, et al. FoxOs are J Clin 2012;62:10–29. lineage-restricted redundant tumor suppressors and regulate endo- 2. Hannenhalli S, Kaestner KH. The evolution of Fox genes and their role thelial cell homeostasis. Cell 2007;128:309–23. in development and disease. Nat Rev Genet 2009;10:233–40. 5. Tothova Z, Kollipara R, Huntly BJ, Lee BH, Castrillon DH, Cullen DE, 3. Carter ME, Brunet A. FOXO transcription factors. Curr Biol 2007;17: et al. FoxOs are critical mediators of hematopoietic stem cell resis- R113–4. tance to physiologic oxidative stress. Cell 2007;128:325–39.

www.aacrjournals.org Cancer Res; 73(17) September 1, 2013 OF9

Zhang et al.

6. Wang Z, Ahmad A, Li Y, Banerjee S, Kong D, Sarkar FH. Forkhead box 25. Pei H, Li L, Fridley BL, Jenkins GD, Kalari KR, Lingle W, et al. FKBP51 M1 transcription factor: a novel target for cancer therapy. Cancer Treat affects cancer cell response to chemotherapy by negatively regulating Rev 2009;36:151–6. Akt. Cancer Cell 2009;16:259–66. 7. Xia JT, Wang H, Liang LJ, Peng BG, Wu ZF, Chen LZ, et al. Over- 26. Grutzmann R, Pilarsky C, Ammerpohl O, Luttges J, Bohme A, Sipos B, et al. expression of FOXM1 is associated with poor prognosis and clinico- Gene expression profiling of microdissected pancreatic ductal carci- pathologic stage of pancreatic ductal adenocarcinoma. Pancreas nomas using high-density DNA microarrays. Neoplasia 2004;6:611–22. 2012;41:629–35. 27. Prasad NB, Biankin AV, Fukushima N, Maitra A, Dhara S, Elkahloun AG, 8. Wang Z, Banerjee S, Kong D, Li Y, Sarkar FH. Down-regulation of et al. Gene expression profiles in pancreatic intraepithelial neoplasia Forkhead Box M1 transcription factor leads to the inhibition of invasion reflect the effects of Hedgehog signaling on pancreatic ductal epithe- and angiogenesis of pancreatic cancer cells. Cancer Res 2007;67: lial cells. Cancer Res 2005;65:1619–26. 8293–300. 28. Browne G, Sayan AE, Tulchinsky E. ZEB proteins link cell motility with 9. Kaestner KH, Silberg DG, Traber PG, Schutz G. The mesenchymal cell cycle control and cell survival in cancer. Cell Cycle 2010;9:886–91. winged helix transcription factor Fkh6 is required for the control of 29. Spaderna S, Schmalhofer O, Wahlbuhl M, Dimmler A, Bauer K, Sultan gastrointestinal proliferation and differentiation. Genes Dev 1997;11: A, et al. The transcriptional repressor ZEB1 promotes metastasis and 1583–95. loss of cell polarity in cancer. Cancer Res 2008;68:537–44. 10. Perreault N, Katz JP, Sackett SD, Kaestner KH. Foxl1 controls the Wnt/ 30. Brabletz S, Brabletz T. The ZEB/miR-200 feedback loop–a motor of beta-catenin pathway by modulating the expression of proteoglycans cellular plasticity in development and cancer? EMBO Rep 2010;11: in the gut. J Biol Chem 2001;276:43328–33. 670–7. 11. Perreault N, Sackett SD, Katz JP, Furth EE, Kaestner KH. Foxl1 is a 31. Fu Z, Tindall DJ. FOXOs, cancer and regulation of apoptosis. Onco- mesenchymal Modifier of Min in carcinogenesis of stomach and colon. gene 2008;27:2312–9. Genes Dev 2005;19:311–5. 32. Brunet A, Bonni A, Zigmond MJ, Lin MZ, Juo P, Hu LS, et al. Akt 12. Moller E, Hornick JL, Magnusson L, Veerla S, Domanski HA, Mertens F. promotes cell survival by phosphorylating and inhibiting a Forkhead FUS-CREB3L2/L1-positive sarcomas show a specific gene expres- transcription factor. Cell 1999;96:857–68. sion profile with upregulation of CD24 and FOXL1. Clin Cancer Res 33. Sunters A, Fernandez de Mattos S, Stahl M, Brosens JJ, Zoumpou- 2011;17:2646–56. lidou G, Saunders CA, et al. FoxO3a transcriptional regulation of Bim 13. Aaltonen LA, Hamilton SR, World Health Organization, International controls apoptosis in paclitaxel-treated breast cancer cell lines. J Biol Agency for Research on Cancer. Pathology and genetics of tumours of Chem 2003;278:49795–805. the digestive system. LyonOxford: IARC Press; Oxford University 34. Dijkers PF, Medema RH, Lammers JW, Koenderman L, Coffer PJ. Press (distributor); 2000. Expression of the pro-apoptotic Bcl-2 family member Bim is regulated 14. Glynn SA, Boersma BJ, Dorsey TH, Yi M, Yfantis HG, Ridnour LA, et al. bythe forkheadtranscriptionfactor FKHR-L1.CurrBiol2000;10:1201–4. Increased NOS2 predicts poor survival in estrogen receptor-negative 35. Wiley SR, Schooley K, Smolak PJ, Din WS, Huang CP, Nicholl JK, et al. breast cancer patients. J Clin Invest 2010;120:3843–54. Identification and characterization of a new member of the TNF family 15. Warner SL, Stephens BJ, Nwokenkwo S, Hostetter G, Sugeng A, that induces apoptosis. Immunity 1995;3:673–82. Hidalgo M, et al. Validation of TPX2 as a potential therapeutic target 36. Pitti RM, Marsters SA, Ruppert S, Donahue CJ, Moore A, Ashkenazi A. in pancreatic cancer cells. Clin Cancer Res 2009;15:6519–28. Induction of apoptosis by Apo-2 ligand, a new member of the tumor 16. Modur V, Nagarajan R, Evers BM, Milbrandt J. FOXO proteins regulate necrosis factor cytokine family. J Biol Chem 1996;271:12687–90. tumor necrosis factor-related apoptosis inducing ligand expression. 37. Pan G, O'Rourke K, Chinnaiyan AM, Gentz R, Ebner R, Ni J, et al. The Implications for PTEN mutation in prostate cancer. J Biol Chem receptor for the cytotoxic ligand TRAIL. Science 1997;276:111–3. 2002;277:47928–37. 38. Pan G, Ni J, Wei YF, Yu G, Gentz R, Dixit VM. An antagonist decoy 17. Aigner K, Dampier B, Descovich L, Mikula M, Sultan A, Schreiber M, receptor and a death domain-containing receptor for TRAIL. Science et al. The transcription factor ZEB1 (deltaEF1) promotes tumour cell 1997;277:815–8. dedifferentiation by repressing master regulators of epithelial polarity. 39. MacFarlane M, Ahmad M, Srinivasula SM, Fernandes-Alnemri T, Oncogene 2007;26:6979–88. Cohen GM, Alnemri ES. Identification and molecular cloning of two 18. Wellner U, Schubert J, Burk UC, Schmalhofer O, Zhu F, Sonntag A, novel receptors for the cytotoxic ligand TRAIL. J Biol Chem et al. The EMT-activator ZEB1 promotes tumorigenicity by repressing 1997;272:25417–20. stemness-inhibiting microRNAs. Nat Cell Biol 2009;11:1487–95. 40. Ashkenazi A. Targeting death and decoy receptors of the tumour- 19. Agnelli L, Mosca L, Fabris S, Lionetti M, Andronache A, Kwee I, et al. A necrosis factor superfamily. Nat Rev Cancer 2002;2:420–30. SNP microarray and FISH-based procedure to detect allelic imbal- 41. LeBlanc HN, Ashkenazi A. Apo2L/TRAIL and its death and decoy ances in multiple myeloma: an integrated genomics approach reveals receptors. Cell Death Differ 2003;10:66–75. a wide gene dosage effect. Genes Chromosomes Cancer 2009;48: 42. Leverkus M, Neumann M, Mengling T, Rauch CT, Brocker EB, Kram- 603–14. mer PH, et al. Regulation of tumor necrosis factor-related apoptosis- 20. Carrasco DR, Tonon G, Huang Y, Zhang Y, Sinha R, Feng B, et al. High- inducing ligand sensitivity in primary and transformed human kerati- resolution genomic profiles define distinct clinico-pathogenetic sub- nocytes. Cancer Res 2000;60:553–9. groups of multiple myeloma patients. Cancer Cell 2006;9:313–25. 43. Gazitt Y. TRAIL is a potent inducer of apoptosis in myeloma cells 21. Zhan F, Barlogie B, Arzoumanian V, Huang Y, Williams DR, Hollmig K, derived from multiple myeloma patients and is not cytotoxic to et al. Gene-expression signature of benign monoclonal gammopathy hematopoietic stem cells. Leukemia 1999;13:1817–24. evident in multiple myeloma is linked to good prognosis. Blood 44. Satoh K, Kaneko K, Hirota M, Masamune A, Satoh A, Shimosegawa T. 2007;109:1692–700. Tumor necrosis factor-related apoptosis-inducing ligand and its 22. Badea L, Herlea V, Dima SO, Dumitrascu T, Popescu I. Combined gene receptor expression and the pathway of apoptosis in human pancre- expression analysis of whole-tissue and microdissected pancreatic atic cancer. Pancreas 2001;23:251–8. ductal adenocarcinoma identifies genes specifically overexpressed in 45. Kim K, Fisher MJ, Xu SQ, el-Deiry WS. Molecular determinants of tumor epithelia. Hepatogastroenterology 2008;55:2016–27. response to TRAIL in killing of normal and cancer cells. Clin Cancer Res 23. Segara D, Biankin AV, Kench JG, Langusch CC, Dawson AC, Skalicky 2000;6:335–46. DA, et al. Expression of HOXB2, a retinoic acid signaling target in 46. Satoh K, Shimosegawa T, Masamune A, Hirota M, Koizumi M, Toyota pancreatic cancer and pancreatic intraepithelial neoplasia. Clin Cancer T. Fas ligand is frequently expressed in human pancreatic duct cell Res 2005;11:3587–96. carcinoma. Pancreas 1999;19:339–45. 24. Ishikawa M, Yoshida K, Yamashita Y, Ota J, Takada S, Kisanuki H, et al. 47. Ungefroren H, Voss M, Jansen M, Roeder C, Henne-Bruns D, Kremer Experimental trial for diagnosis of pancreatic ductal carcinoma based B, et al. Human pancreatic adenocarcinomas express Fas and Fas on gene expression profiles of pancreatic ductal cells. Cancer Sci ligand yet are resistant to Fas-mediated apoptosis. Cancer Res 2005;96:387–93. 1998;58:1741–9.

OF10 Cancer Res; 73(17) September 1, 2013 Cancer Research

FOXL1, a Novel Candidate Tumor Suppressor, Inhibits Tumor Aggressiveness and Predicts Outcome in Human Pancreatic Cancer

Geng Zhang, Peijun He, Jochen Gaedcke, et al.

Cancer Res Published OnlineFirst June 25, 2013.

Updated version Access the most recent version of this article at: doi:10.1158/0008-5472.CAN-13-0362

Supplementary Access the most recent supplemental material at: Material http://cancerres.aacrjournals.org/content/suppl/2013/06/25/0008-5472.CAN-13-0362.DC1

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://cancerres.aacrjournals.org/content/early/2013/08/23/0008-5472.CAN-13-0362. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.