[ RESEARCH 64, 7711–7723, November 1, 2004] Receptor Regulates Aberrant Expression of Insulin-Like Growth Factor-Binding 3

Munenori Takaoka,1 Hideki Harada,1 Claudia D. Andl,1 Kenji Oyama,1 Yoshio Naomoto,2 Kelly L. Dempsey,1 Andres J. Klein-Szanto,3 Wafik S. El-Deiry,4,5,6 Adda Grimberg,7 and Hiroshi Nakagawa1 1Gastroenterology Division, Department of Medicine, Abramson Cancer Center, University of Pennsylvania, Philadelphia, Pennsylvania; 2Department of Gastroenterological Surgery, Transplant, and Surgical Oncology, Graduate School of Medicine and Dentistry, Okayama University, Okayama, Japan; 3Department of Pathology, Fox Chase Cancer Center, Philadelphia, Pennsylvania; Departments of 4Medicine, 5Genetics, and 6Pharmacology, University of Pennsylvania, Philadelphia, Pennsylvania; and 7Pediatric Endocrinology, Children’s Hospital of Philadelphia, University of Pennsylvania, Philadelphia, Pennsylvania

ABSTRACT mal growth factor receptor (EGFR) and cyclin D1 overexpression, inactivation of and p16INK4a, and telomerase activation (1). Epidermal growth factor receptor (EGFR) is frequently overexpressed EGFR is a receptor tyrosine kinase whose biochemical and struc- in esophageal carcinoma and its precursor lesions. To gain insights into tural properties have been extensively studied (2). EGFR critically how EGFR overexpression affects cellular functions in primary human regulates a number of cellular functions, including cell proliferation, esophageal cells, we performed profiling and identified insulin-like growth factor-binding protein (IGFBP)-3 as the most up- survival, differentiation, migration, cell–cell adhesion, and cell– regulated gene. IGFBP-3 regulates cell proliferation through both insulin- extracellular matrix interactions. EGFR is a classical proto-oncogene like growth factor-dependent and independent mechanisms. We found that can transform NIH3T3 fibroblasts in a ligand-dependent fashion that IGFBP-3 mRNA and protein expression was increased in EGFR- (3). EGFR overexpression, partially accounted for by gene amplifi- overexpressing primary and immortalized human esophageal cells. IG- cation, is found in up to 80% of esophageal squamous cell carcinomas FBP-3 was also up-regulated in EGFR-overexpressing cells in organotypic (ESCCs) and esophageal adenocarcinomas (EACs), as well as their culture and in EGFR transgenic mice. Furthermore, IGFBP-3 mRNA was precursor lesions, such as squamous dysplasia and Barrett’s esopha- overexpressed in 80% of primary esophageal squamous cell carcinomas gus, respectively (4, 5). However, most studies on EGFR biological and 60% of primary esophageal adenocarcinomas. Concomitant up- functions have been carried out in rodent fibroblasts or human cancer regulation of EGFR and IGFBP-3 was observed in 60% of primary cell lines, thus hindering insights into the molecular and physiologic esophageal squamous cell carcinomas. Immunohistochemistry revealed mechanisms through which EGFR overexpression contributes to im- cytoplasmic localization of IGFBP-3 in the preponderance of preneoplas- mortalization and malignant transformation. tic and neoplastic esophageal lesions. IGFBP-3 was also overexpressed in esophageal cancer cell lines at both mRNA (60%) and protein (40%) We have recently isolated and characterized primary human esoph- levels. IGFBP-3 secreted by cancer cells was capable of binding to insulin- ageal cells (6). Moreover, using retrovirus-mediated transduction, we like growth factor I. Functionally, epidermal growth factor appeared to established stable cell lines expressing EGFR and/or the catalytic regulate IGFBP-3 expression in esophageal cancer cell lines. Finally, subunit of human telomerase (hTERT; refs. 6, 7). Primary esophageal suppression of IGFBP-3 by small interfering RNA augmented cell prolif- cells overexpressing EGFR demonstrated unique cell biological prop- eration, suggesting that IGFBP-3 may inhibit tumor cell proliferation as a erties, including increased migration associated with induction of negative feedback mechanism. In aggregate, we have identified for the matrix metalloproteinase-1, pronounced cell aggregation through en- first time that IGFBP-3 is an aberrantly regulated gene through the hanced functions of E-cadherin, and basal cell hyperplasia in organo- EGFR signaling pathway and it may modulate EGFR effects during typic culture (6). However, EGFR alone failed to immortalize primary carcinogenesis. esophageal cells. By contrast, hTERT-mediated telomerase activation resulted in immortalization of primary human esophageal cells with- out affecting the p53 and pRb pathways (7). INTRODUCTION Insulin-like growth factor-binding protein (IGFBP)-3, one of six Among , esophageal carcinoma is one of the leading causes IGFBPs, is the major carrier protein for insulin-like growth factor of cancer-related mortality worldwide. Esophageal malignancies are (IGF)-I or IGF-II in circulation. IGFBP-3 protein has molecular predominantly of two types: squamous cell carcinoma and adenocar- masses of 43 to 45 kDa, depending on posttranslational modifications cinoma. Etiologically, the former is associated with cigarette smoking such as glycosylation and phosphorylation. It exists as a component of and alcohol, whereas the latter is associated with Barrett’s esophagus, a 150-kDa ternary complex comprising an 85-kDa acid labile glyco- which is characterized by intestinal metaplasia of the esophageal protein subunit and IGF-I or IGF-II (8, 9). Various cell types includ- squamous epithelium caused in part by acid reflux. Common genetic ing fibroblasts, endothelial cells, and epithelial cells secrete IGFBP-3 lesions identified in both forms of esophageal cancer include epider- (10–12). Although many endogenous and pharmacological agents, including peptide growth factors, cytokines, and hormones, have been shown to induce IGFBP-3, how IGFBP-3 expression is regulated Received 2/26/04; revised 7/26/04; accepted 8/25/04. Grant support: National Institutes of Health grant R21 DK64249 and American remains largely undetermined. Gastroenterological Association/Foundation for Digestive Health and Nutrition Research IGFBP-3 has been demonstrated to regulate cell proliferation Scholar Award (H. Nakagawa); National Institute of Diabetes and Digestive and Kidney Diseases grant K08 DK64352 (A. Grimberg); National Cancer Institute grant P01- through both IGF-dependent and independent mechanisms (9, 13). CA098101 (H. Nakagawa, C. Andl, K. Dempsey, W. El-Deiry, and A. Klein-Szanto) and IGFBP-3 is known to be up-regulated in senescent cells or mitotically its core facilities; and National Institutes of Health/National Institute of Diabetes and quiescent cells (11, 12, 14). However, little knowledge is available Digestive and Kidney Diseases Center for Molecular Studies in Digestive and Liver Diseases and its Morphology Core, Molecular Biology, and Core facilities; regarding IGFBP-3 expression and function in cancer. Depending on and pilot grant program P30 DK50306. the experimental context, IGFBP-3 has been shown to possess The costs of publication of this article were defrayed in part by the payment of page growth-stimulatory, antiproliferative, or proapoptotic activities in charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. vitro. In transgenic mice, ubiquitous expression of IGFBP-3 resulted Requests for reprints: Hiroshi Nakagawa, Gastroenterology Division, University of in growth retardation, suggesting that IGFBP-3 may impair the Pennsylvania, 415 Curie Boulevard, 638 CRB, Philadelphia, PA 19104-2144. Phone: 215-573-1867; Fax: 215-573-2024; E-mail: [email protected]. growth-promoting functions of IGFs in vivo (15). However, tissue- ©2004 American Association for Cancer Research. specific targeted expression of the IGFBP-3 transgene by metallothio- 7711 IGFBP-3 OVEREXPRESSION IN ESOPHAGEAL CANCER nein-I induced selective organomegaly of the spleen, liver, and heart sulfoxide (Sigma, St. Louis, MO) and added into culture medium at the (16). In other IGFBP-3 transgenic mice, the mammary gland under- indicated concentrations to inhibit EGFR-tyrosine kinase activity. went delayed involution after lactation, suggesting that IGFBP-3 Small Interfering RNA Preparation. To stably express short hairpin overexpression potentiated the antiapoptotic activities mediated by small interfering RNA (siRNA) under the control of the H1-RNA polymerase Ј IGFs (17). III promoter, complementary oligonucleotides consisting of 5 -gatccccATGCTA- GTGAGTCGGAGGAttcaagagaTCCTCCGACTCACTAGCATtttttggaaa-3Ј and To gain further insights into how EGFR overexpression affects 5Ј-agcttttccaaaaaATGCTAGTGAGTCGGAGGAtctcttgaaTCCTCCGACTCAC- cellular functions of primary human esophageal cells, we carried out TAGCATggg-3Ј, including sense and antisense 19-nucleotide sequences specific gene expression profiling and identified IGFBP-3 as a critically up- for human IGFBP-3 mRNA (Genebank sequence X64875; nucleotides 539–557), regulated gene. We find that IGFBP-3 is frequently overexpressed in separated by a 9-nucleotide spacer, were annealed, and subcloned into the BglII esophageal cancer cell lines and primary esophageal tumors. We also and HindIII sites of the pSuper-retro-neo vector (Oligoengine, Seattle, WA), demonstrate a basis for IGFBP-3 regulation by EGFR. Moreover, resulting in the creation of pSRN-IGFBP-3-3, according to the manufacturer’s IGFBP-3 appears to negatively regulate cell proliferation in esopha- instruction. geal cancer cells. Indeed, the novel findings described herein may Retrovirus-Mediated Transduction. Stable transduction of primary help to explain how EGFR through IGFBP-3 may modulate cell esophageal cells with retroviral vectors was described previously (6, 7). The pFB-neo retroviral vector (Stratagene, La Jolla, CA) containing the entire growth but not induce immortalization and malignant transformation. coding sequence for the human EGFR (pFB-neo-WT-hEGFR) or green fluo- rescent protein [GFP (pFB-neo-GFP)], pSuper-retro-neo, and pSRN-IGFBP- 3-3 were transfected into a Phoenix-Ampho packaging cell line (gift of Dr. MATERIALS AND METHODS Garry Nolan, Stanford University, Palo Alto, CA) with LipofectAMINE 2000 reagent (Invitrogen), according to the manufacturer’s instructions. Culture Tissue Samples. Esophageal tissue samples were procured via surgery supernatants from individual Phoenix-Ampho cells were used to infect EPC1, from the Okayama University Hospital (21 cases) and the Hospital of the EPC2, and EPC2-hTERT as well as TE11 cells. Cells were passaged 48 hours University of Pennsylvania through the Cooperative Human Tissue Network after infection and selected with 300 ␮g/mL Geneticin (Invitrogen) for 7 days, (11 cases). Eight and 24 cases were pathologically diagnosed as EAC and resulting in generation of EPC2-EGFR, EPC2-GFP, EPC2-Neo, EPC2- ESCC, respectively. All of the ESCC tissues contained lesions consistent with hTERT-EGFR, and EPC2-hTERT-Neo. TE11 cells transduced with the siRNA squamous dysplasia. Paraffin blocks representing tumors and adjacent normal vectors were also similarly selected with 300 ␮g/mL Geneticin. tissues were available from 6 of 8 EAC cases and 17 of 24 ESCC cases for RNA Isolation and cDNA Synthesis. TRIzol reagent (Invitrogen) and the immunohistochemistry. Paired frozen tissues representing tumors and adjacent RNeasy Fibrous Tissue Midi Kit (Qiagen, Valencia, CA) were used to isolate normal tissues were available from 7 of 8 EAC cases and 19 of 24 ESCC cases total RNA from cultured cells and snap-frozen tissues, respectively. RNA was for RNA or protein analyses. All of the clinical materials were obtained from treated with DNase I (Invitrogen). The integrity of total RNA was determined patients who provided informed consent in accordance with institutional re- by 1% formaldehyde-agarose gel electrophoresis. cDNA synthesis was carried view board standards and guidelines. out with the Superscript First Strand Synthesis System (Invitrogen) using 3.3 EGFR Transgenic Mice. Generation of EGFR transgenic mice was de- (cell RNA) or 5 ␮g (tissue RNA) of total RNA as a template. The cDNA scribed previously (6). The EGFR transgene was targeted in the oral-esopha- synthesis reactions without reverse transcriptase yielded no amplicons in the geal epithelium in a tissue-specific fashion with the Epstein-Barr virus ED-L2 polymerase chain reaction (PCR) reactions described below. promoter (18). The EGFR transgenic mice and age-matched wild-type mice Gene Expression Profiling. Gene expression profiling of EPC2-EGFR were sacrificed at age 6 months for immunohistochemistry. All transgenic and EPC2-GFP cells was carried out using the Affymetrix HG-U95A Human mouse experiments were carried out in accordance with the standards and GeneChip array (Affymetrix, Santa Clara, CA). Preparation of cRNA, hybrid- guidelines of the Institutional Animal Care and Use Committee at the Univer- ization, and scanning of the arrays were performed according to manufactur- sity of Pennsylvania. ␮ er’s protocols. Briefly, 5 g of total RNA were primed by an oligo(dT)24 Cell Lines. Fifteen human esophageal carcinoma cell lines (TE series, T.T., primer containing a T7 RNA polymerase promoter 3Ј to the poly(T) (Geneset, HCE4, and HCE7) and six nonesophageal cell lines (HepG2, PANC-1, MCF7, La Jolla, CA) and converted to first-strand cDNA using Superscript II reverse HeLa, A431, and HaCaT) were cultured under standard conditions as de- transcriptase (Invitrogen). Second-strand cDNA synthesis was followed by in scribed previously (19). TE7 cells were derived from adenocarcinoma, vitro transcription for linear amplification of each transcript incorporating whereas the other esophageal carcinoma cell lines were established from biotin-11-CTP and biotin-16-UTP (Enzo, Farmington, NY) to generate labeled ESCCs. EPC1 and EPC2 are primary human esophageal keratinocytes that cRNA. The labeled cRNA was purified over RNeasy columns. Fifteen micro- have been described previously (6, 7). EPC1 and EPC2 cells were grown at grams of cRNA were fragmented to Յ200 nucleotides at 94°C for 35 minutes 37°C under 5% CO2 in serum-free medium (Keratinocyte-SFM) supplemented in 40 mmol/L Tris-acetate (pH 8.1), 100 mmol/L potassium acetate, and 30 with 40 ␮g/mL bovine pituitary extract and 1 ng/mL epidermal growth factor mmol/L magnesium acetate; heat-denatured at 99°C for 5 minutes; and hy- (EGF; Invitrogen, Carlsbad, CA). Esophageal epithelial cells (designated bridized for 16 hours at 45°C to the HG-U95A microarray in 200 ␮Lof mouse esophageal keratinocytes) were isolated from the EGFR transgenic hybridization mixture consisting of 100 mmol/L 4-morpholineethanesulfonic mouse and the control mouse and grown in culture as described previously acid, 1 mol/L NaCl, 20 mmol/L EDTA, 0.01% Tween 20, 0.1 mg/mL herring (20). Cells were counted using Coulter Z1 counter (Beckman Coulter Inc., sperm DNA (Promega, Madison, WI), and 500 ␮g/mL acetylated bovine Fullerton, CA). Cell viability was determined by trypan blue exclusion. Con- serum albumin (Invitrogen). The microarray was then washed at low (0.9 ditioned medium (CM) was harvested by incubating subconfluent cells for an mol/L NaCl, 60 mmol/L NaH2PO4, 6 mmol/L EDTA, and 0.01% Tween 20) indicated time period with full Keratinocyte-SFM or Dulbecco’s modified and high (100 mmol/L 4-morpholineethanesulfonic acid, 0.1 mol/L NaCl, and Eagle’s medium (DMEM) containing either 10% fetal calf serum (FCS) or 0.01% Tween 20) stringency and stained with streptavidin-phycoerythrin (Mo- 0.5% FCS. To concentrate CM, Centricon YM-10 centrifugal filter units lecular Probes, Eugene, OR). Fluorescence was amplified by adding biotiny- (10,000 nominal molecular weight limit; Millipore, Billerica, MA) were used lated antistreptavidin and an additional aliquot of streptavidin-phycoerythrin according to the manufacturer’s instructions. stain. The GeneArray scanner (Affymetrix) was used to collect fluorescence Organotypic cell culture was performed as described previously (6). In signal at 3 ␮m resolution after excitation at 570 nm. The average signal from brief, 5 ϫ 105 esophageal cells were seeded onto collagen matrix containing two sequential scans was calculated for each microarray feature. The Af- 7.5 ϫ 104 human skin fibroblast cells. After cell culture under submerged fymetrix array data were analyzed using Affymetrix Microarray Suite 5.0 conditions for 4 days, cells were exposed to the air–liquid interface and (Affymetrix) and Genespring 5.0 (Silicon Genetics, Redwood City, CA), cultured for an additional 6 days. The reconstituted epithelial sheets and the programs designed for the analysis of high-throughput microarray expression collagen matrix were fixed with 10% formaldehyde, embedded in paraffin, and data. Genes that are absent in all samples were removed from consideration. subjected to histology and immunohistochemistry. The list of genes was condensed to genes that changed Ն3-fold (whether up or AG1478 (Calbiochem, San Diego, CA) was dissolved in 0.1% dimethyl down) between EPC2-GFP and EPC2-EGFR cells. 7712 IGFBP-3 OVEREXPRESSION IN ESOPHAGEAL CANCER

Real-Time Reverse Transcription-Polymerase Chain Reaction. Real- was visualized by an enhanced chemiluminescence solution (ECL Plus; Am- time PCR was performed using SYBR green reagent (PE Applied Biosystems, ersham Pharmacia Biotech) and exposed to Eastman Kodak Co. (Rochester, Foster City, CA) and the ABI PRISM 7000 Sequence Detection System (PE NY) X-OMAT LS film. Western blots were quantified by densitometry with Applied Biosystems) according to the manufacturer’s instructions. The fol- Scion Image ␤ 4.02 program (Scion Corp., Frederick, MD). lowing oligonucleotides were used as primers: (a) human IGFBP-3, 5Ј-CCAT- Western Ligand Blotting. Western ligand blotting was done as described GACTGAGGAAAGGAGCTC-3Ј (BP3–2218-F; exon 4; nucleotides 10487– previously (21), with modifications. Briefly, protein samples were electro- 10508; forward primer) and 5Ј-TGCAGCAGGGCAGAGTCTC-3Ј (BP3– phoresed on a NuPAGE Bis-Tris 4% to 12% gel under a nonreduced condition 2325-R; exon 4, nucleotides 10576–10594; GenBank accession no. M35878; and transferred to a polyvinylidene difluoride membrane (Immobilon-P). The reverse primer); (b) mouse IGFBP-3, 5Ј-GAGTGTGGAAAGCCAGGTT- membrane was rinsed once with TBS containing 3% Nonidet P-40 for 30 GTC-3Ј (mBP3525-F; forward primer) and 5Ј-GCATGGAGTGGATG- minutes at 4°C and then rinsed once with TBST containing 1% bovine serum GAACTTG-3Ј (mBP3593-R; GenBank accession no. NM_008343; reverse albumin for 2 hours at 4°C. The membrane was incubated with 1 ␮Ci of primer); (c) human ␤-actin, 5Ј-CCTGGCACCCAGCACAAT-3Ј (exon 5; 3-[125I]iodotyrosyl recombinant human IGF-I (Amersham Pharmacia Biotech) hBAC-F; forward primer) and 5Ј-GCCGATCCACACGGAGTACT-3Ј (exon in 10 mL of TBST containing 1% bovine serum albumin for 16 hours at 4°C. 6; hBAC-R; reverse primer); (d) human glyceraldehyde-3-phosphate dehydro- After a brief wash with TBST, the membrane was subjected to autoradiography genase (GAPDH), 5Ј-GGTGGTCTCCTCTGACTTCAACA-3Ј (exon 7; using Kodak X-OMAT AR film with an intensifying screen at Ϫ80°C. hGAPDH-F; forward primer) and 5Ј-GTTGCTGTAGCCAAATTCGTTGT-3Ј Enzyme-Linked Immunosorbent Assay. Enzyme-linked immunosorbent (exon 8; hGAPDH-R; reverse primer); and (e) mouse GAPDH, 5Ј-GGTT- assay (ELISA) was performed using human IGFBP-3 DuoSet ELISA devel- GTCTCCTGCGACTTCAAC-3Ј (mGAPDH-F; forward primer) and 5Ј- opment systems (R&D Systems Inc., Minneapolis, MN), according to the CCAGGAAATGAGCTTGACAAAGTT-3Ј (mGAPDH-R; reverse primer). manufacturer’s instructions. Briefly, a BD Falcon 96-well microplate (BD ␤-Actin was used as an internal control for human tissues. GAPDH was used Biosciences, San Jose, CA) was coated with mouse antihuman IGFBP-3 as an internal control for cultured cells as well as mouse tissues. PCR capture antibody and blocked with 5% Tween 20 and 5% sucrose in PBS

conditions were optimized by performing primer matrix reactions and gener- containing 0.05% NaN3 at room temperature for 16 hours. After washing with ating standard curves for IGFBP-3, ␤-actin, and GAPDH. All PCR reactions wash buffer consisting of 0.05% Tween 20 and 5% sucrose in PBS [137 ␮ were performed in triplicate in a 25- L total volume using 500 nmol/L each of mmol/L NaCl, 2.7 mmol/L KCl, 8.1 mmol/L Na2HPO4, and 1.5 mmol/L ␮ forward and reverse primers. One microliter of cDNA was used as template per KH2PO4 (pH 7.4)] containing 0.05% NaN3, 100 L of CM samples or serially reaction to detect human and mouse IGFBP-3 mRNA, whereas 0.125 ␮Lof diluted recombinant human IGFBP-3 were added into each well and incubated cDNA was used for ␤-actin and GAPDH. The relative expression level of for 2 hours at room temperature and washed three times with the wash buffer. IGFBP-3 mRNA between normal and tumor was calculated by normalizing to Signal was visualized by incubation with 100 ␮L/well of 36 ␮g/mL biotiny- ␤ ⌬⌬ ␮ -actin mRNA expression level using the comparative CT ( CT) method, lated goat antihuman IGFBP-3 antibody, 100 L/well streptavidin-horseradish ␮ where CT represents the cycle number at which the amplification reaches a peroxidase, and 100 L/well of a 1:1 mixture of H2O2 and tetramethylbenzi- ␮ threshold level chosen to lie in the exponential phase of all PCR reactions. dine. The reaction was terminated by 50 L/well of 2N H2SO4, and the Ϫ⌬⌬ Relative expression levels were calculated using the formula 2 CT, where absorbance of each well was immediately determined using the Vmax kinetic ⌬ CT represents the difference between the average IGFBP-3 CT value and the microplate reader (Molecular Devices Corp., Sunnyvale, CA) at 450 nm with ␤ ⌬⌬ average -actin CT value within a given tissue. CT represents the difference correction at 546 nm. ⌬ between the CT values for normal and tumor. The normalized IGFBP-3 Immunohistochemistry. Immunohistochemistry for IGFBP-3 was per- expression level for normal tissues was set to 1. Likewise, the relative expres- formed with the Vectastain Elite kit (Vector Laboratories, Burlingame, CA) sion level of IGFBP-3 mRNA among cell lines was determined by normalizing following the manufacturer’s protocol. In brief, paraffin sections were depar- to the GAPDH mRNA level. EPC2 [3 population doublings (PDs)] was used affinized with xylene, hydrated in descending ethanol solutions, and then as a normal control. Thus, the normalized IGFBP-3 mRNA expression level in placed in a microwave in 10 mmol/L citric acid buffer. Endogenous peroxidase EPC2 (3 PDs) was set to 1. Data were analyzed using ABI PRISM 7000 was quenched using hydrogen peroxide before sections were blocked in avidin sequence detection system software (PE Applied Biosystems). D blocking reagent and biotin blocking reagent. Sections were incubated with Antibodies. Affinity-purified goat antihuman IGFBP-3 (DSL-R00536; Di- primary antihuman IGFBP-3 mouse monoclonal antibody (clone 84728.111; agnostic Systems Laboratories, Inc., Webster, TX), affinity-purified rat anti- R&D Systems Inc.) at a 1:250 dilution and with biotinylated antimouse IgG at mouse IGFBP-3 (138202; R&D Systems Inc., Minneapolis, MN), affinity- a 1:200 dilution, and then signal was developed using the 3,3Ј-diaminobenzi- purified goat antihuman IGFBP-4 (DSL-R00637; Diagnostic Systems dine substrate kit for peroxidase. The immunohistochemical staining was Laboratories, Inc.), anti-EGFR Ab-12 mouse monoclonal antibody (Cocktail assessed independently (A. J. K-S.), and the intensity was expressed as nega- R19/48; NeoMarkers, Union City, CA), anti–␤-actin (Sigma), anti-tubulin tive (Ϫ), weakly positive (ϩ), moderately positive (ϩϩ), or strongly positive (DMA1AϩDM1B; Neo Markers), and secondary horseradish peroxidase- (ϩϩϩ). conjugated donkey antigoat IgG (sc-2020; Santa Cruz Biotechnology, Santa Immunocytochemistry. Immunocytochemistry was performed as de- Cruz, CA) and sheep antimouse IgG (Amersham Pharmacia Biotech, Piscat- scribed previously (6). Briefly, cells grown in chamber slides (Nalge Nunc, away, NJ) were purchased. Inc., Naperville, IL) were fixed in 1:1 methanol/acetone for 10 minutes at Western Blotting. Western blotting was carried out as described previ- Ϫ20°C. After blocking with 1% bovine serum albumin (Sigma) for 10 min- ously (6). In brief, cells and tissues were homogenized with a radioimmuno- utes, slides were incubated with antihuman IGFBP-3 mouse monoclonal an- precipitation assay buffer consisting of 10 mmol/L Tris-HCl (pH 7.4), 150 tibody (clone 84728.111; R&D Systems Inc.) at a 1:250 dilution overnight at mmol/L NaCl, 1% Nonidet P-40, 0.1% SDS, 0.1% sodium deoxycholate, 1 4°C and then incubated with Cy3-conjugated secondary donkey antimouse IgG

mmol/L EDTA, 2 mmol/L Na3VO4, 1 mmol/L phenylmethylsulfonyl fluoride, antibody (Jackson ImmunoResearch Laboratories, Inc., West Grove, PA) at a and a protease inhibitor mixture tablet (Complete; Roche Applied Science, 1:400 dilution for 30 minutes at 37°C. Stained slides were examined with a Indianapolis, IN) and cleared by centrifugation at 14,000 rpm at 4°C for 15 Nikon Microphot microscope and imaged with a digital camera at specific minutes. The total protein sample (20 ␮g) was denatured and fractionated on magnifications. a NuPAGE Bis-Tris 4% to 12% gel with NuPAGE MOPS running buffer using [3H]Thymidine Incorporation Assay. To assess cell proliferation, DNA the NuPAGE system (Invitrogen) and electrotransferred to polyvinylidene synthesis was measured by incubating cells with 1 ␮Ci of [methyl-3H]thymi- difluoride membrane (Immobilon-P; Millipore Corp., Bedford, MA) according dine (1 ␮Ci/mL/well; Perkin-Elmer Life Sciences Inc., Boston, MA) for 4 to the manufacturer’s instructions. After blocking with 5% nonfat milk (Bio- hours in 12-well tissue culture dishes. Cells were washed three times with cold Rad) in Tris-buffered saline [TBS (10 mmol/L Tris, 150 mmol/L NaCl, pH PBS, washed once with cold 10% trichloroacetic acid, and washed three times 8.0)] for 1 hour at room temperature, the membranes were probed with primary with 5% trichloroacetic acid. Cells were lysed with 0.5N NaOH on ice for 10 antibody diluted 1:1,000 in TBS with 0.1% Tween 20 (TBST) containing 2% minutes and neutralized with 0.5N HCl. The cell lysate was supplemented with milk overnight at 4°C, washed three times in TBST, incubated with secondary 10% trichloroacetic acid, incubated on ice for 20 minutes, and filtrated with antigoat (1:10,000) or antimouse (1:10,000) antibody diluted in TBST for 1 glass microfiber filters (Whatman, Kent, United Kingdom) using a sampling hour at room temperature, and then washed three times in TBST. The signal vacuum manifold (Millipore, Billerica, MA). The filters were rinsed three 7713 IGFBP-3 OVEREXPRESSION IN ESOPHAGEAL CANCER times with ethanol, dried at 80°C for 45 minutes, and suspended in 10 mL of carried out. EPC2-EGFR cells expressed 4.4-fold more IGFBP-3 Ready Value Liquid Scintillation Cocktail (Beckman Coulter Inc.) to measure mRNA than control cells (Fig. 1A). This was corroborated by North- radioactivity with a LS 6500 Multi-Purpose Scintillation Counter (Beckman ern blotting (data not shown). Western blotting also revealed a 3.5- Coulter Inc.). fold induction of IGFBP-3 protein in EPC2-EGFR cells compared Statistical Analyses. Simple linear regression analysis was performed to with control cells (Fig. 1B). To further confirm IGFBP-3 up-regula- compare EGFR level and IGFBP-3 level between tumors and adjacent normal tion in EGFR-overexpressing cells, retrovirus transduction was car- tissues. Student’s t test was used to compare data between two groups. Data represent the mean Ϯ SE. PsofϽ0.05 were considered statistically significant. ried out in an independent fashion. In EPC2, EPC2-hTERT, and another primary human esophageal line, EPC1, EGFR transduction resulted in IGFBP-3 mRNA and protein induction (Fig. 1A and B). RESULTS We also confirmed by Western blotting that the IGFBP-3 full-length EGFR Induces IGFBP-3 in Primary and hTERT-Immortalized form is secreted in CM from EPC2, EPC2-hTERT, and EPC1 cells Human Esophageal Cells. To identify genes expressed differentially transduced with EGFR (Fig. 1C). An ELISA confirmed that the CM in EPC2-EGFR cells compared with control EPC2-GFP cells, we used from these EGFR-overexpressing cells contained 2.5- to 5-fold higher an Affymetrix HG-U95A Human GeneChip consisting of ϳ12,500 levels of IGFBP-3 than media from control cells (Fig. 1D). Further- probe sets. Two independent data analysis programs, Affymetrix more, immunocytochemistry revealed that both EGFR and IGFBP-3 Microarray Suite 5.0 and Genespring 5.0, extracted 23 and 35 genes, are more intensely expressed in the EGFR-transduced cells than respectively, as Ն3-fold up-regulated in EPC2-EGFR cells. IGFBP-3 control cells (Fig. 1E and F; data not shown). It should be noted that was the most highly up-regulated gene in EPC2-EGFR cells. there was no significant variation in IGFBP-3 expression between To verify the gene array result for IGFBP-3 in EPC2-EGFR cells, cells, even though the primary cell derivatives were pooled after real-time reverse transcription (RT)-PCR and Western blotting were retrovirus infection.

Fig. 1. IGFBP-3 up-regulation and increased secretion into culture media by primary and immortalized human esophageal cells transduced with EGFR. A. IGFBP-3 mRNA was determined by real-time RT-PCR in primary human esophageal cells and their derivatives. Mean Ϯ SE (n ϭ 3) in a representative experiment is shown. B. Protein expression for EGFR and IGFBP-3 was determined in cell lysates by Western blotting. ␤-Actin was used as a loading control. Lane 1, parental EPC2; Lane 2, EPC2-GFP; Lane 3, EPC2-EGFR; Lane 4, EPC2-Neo; Lane 5, EPC2-EGFR; Lane 6, EPC2-hTERT; Lane 7, EPC2-hTERT-Neo; Lane 8, EPC2-hTERT-EGFR; Lane 9, EPC1-Neo; Lane 10, EPC1-EGFR. C. IGFBP-3 secreted by cells into CM was determined by Western blotting using concentrated (10ϫ) CM. D. IGFBP-3 concentration in CM from EPC1, EPC2, and their derivatives was measured by ELISA. For protein analysis, 1 ϫ 106 cells were seeded per 100-mm plate, and medium exchange was done 48 hours before harvesting the CM. Protein yields in cell lysates prepared simultaneously were used to adjust the differences in cell number that may affect IGFBP-3 concentration in CM. Lane 1, EPC2-Neo; Lane 2, EPC2-EGFR; Lane 3, EPC2-hTERT; Lane 4, EPC2-hTERT-Neo; Lane 5, EPC2-hTERT-EGFR; Lane 6, EPC1-Neo; Lane 7, EPC1-EGFR. E and F, immunofluorescence for IGFBP-3 in EPC2-hTERT-EGFR (E) and EPC2-hTERT-neo (F) cells. Magnification, ϫ400. PD, population doublings. 7714 IGFBP-3 OVEREXPRESSION IN ESOPHAGEAL CANCER

IGFBP-3 Is Up-Regulated in EGFR-Overexpressing Reconsti- IGFBP-3 Is Highly Expressed in Primary Esophageal Cancer tuted Esophageal Epithelium and the Esophagus of Transgenic Tissues. To determine IGFBP-3 expression in primary esophageal Mice. To determine the role of EGFR in the induction of IGFBP-3 in cancer tissues, real-time RT-PCR, Western blotting, and immunohis- a tissue context, we examined IGFBP-3 expression in organotypic tochemistry were performed. IGFBP-3 mRNA was overexpressed in culture (also referred to as reconstruct) and ED-L2 EGFR transgenic 15 of 19 (78.9%) primary ESCC tissues, including 8 samples (36.8%) mice. We have recently shown that EGFR overexpression results in that showed a Ն10-fold increase compared with adjacent normal esophageal epithelial hyperplasia in these model systems. In organo- mucosa (Fig. 3A). IGFBP-3 overexpression was also observed in four typic culture, EPC2-EGFR cells formed a taller epithelium than con- of seven EAC tissues (57.1%; Fig. 3B). Western blotting confirmed trol cells with increased basal cell proliferation, which was antago- IGFBP-3 overexpression in ESCC; 8 of 11 ESCC samples showed a nized by AG1478, an EGFR-specific tyrosine kinase inhibitor (6). Ն3-fold increase in IGFBP-3 expression compared with adjacent Immunohistochemistry revealed an extended stratification of IGFBP- normal mucosa (Fig. 3C). The IGFBP-3 mRNA and protein levels 3–positive cells in the EPC2-EGFR epithelium (Fig. 2A), whereas tended to correlate with the EGFR protein level in the tumor tissues IGFBP-3 expression was restricted to the basal compartment of the (R2 ϭ 0.365; P ϭ 0.003). Concomitant up-regulation of EGFR and control cell reconstruct (Fig. 2B). IGFBP-3 was observed in 7 of 11 (60%) samples of primary ESCCs. In the ED-L2 EGFR transgenic mice, real-time RT-PCR on RNA Finally, immunohistochemistry revealed that IGFBP-3 was intensely samples from mouse esophagus also demonstrated a marked induction expressed in tumor tissues of 16 of 17 (94.1%) invasive ESCCs as of IGFBP-3 mRNA compared with wild-type littermates (Fig. 2C). well as 5 of 6 (83.3%) EACs (Fig. 4; data not shown). Whereas When esophageal epithelial cells were isolated from the ED-L2 EGFR IGFBP-3 expression was increased in almost all layers of dysplastic transgenic mice, they maintained EGFR overexpression and IGFBP-3 squamous epithelia (Fig. 4B), it was confined to the basal and supra- induction (Fig. 2D and E). These findings suggest that EGFR induces basal cell layers of the normal esophageal squamous epithelium (Fig. IGFBP-3 in esophageal epithelial cells both in vitro and in vivo. 4A). IGFBP-3 was diffusely distributed in the cytoplasm of both We also observed that IGFBP-3 mRNA was initially down-regu- ESCC and EAC (Fig. 4CϪH). Interestingly, some foci of cancer cells lated when presenescent EPC2 cells (42 PDs) were transduced with had a very intense cytoplasmic staining for IGFBP-3 (Fig. 4H), which hTERT. However, IGFBP-3 expression was increased in immortal- may represent localization of IGFBP-3 before secretion. ized EPC2-hTERT cells at a later PD (Ͼ200 PDs; data not shown). IGFBP-3 Is Functional and Capable of Binding Insulin-Like This prompted us to investigate whether or not IGFBP-3 is up- Growth Factor I. To determine IGFBP-3 expression in esophageal regulated in esophageal cancer cells that are typically associated with cancer cell lines, real-time RT-PCR, Western blotting, and ELISA EGFR overexpression and telomerase activation. were performed. Real-time RT-PCR showed that 9 of 15 (60%)

Fig. 2. IGFBP-3 is up-regulated in primary esophageal cells transduced with EGFR in organotypic culture and EGFR transgenic mouse esophagus. IGFBP-3 immunohistochemistry in the reconstituted epithelium of (A) EPC2-EGFR and (B) EPC2-Neo cells in organotypic culture. Note that IGFBP-3 was also detected in quiescent fibroblasts embedded in the collagen matrix (Col) underlying the reconstituted epithelia (Epi). Magnification: A and B, ϫ200. C. IGFBP-3 mRNA level was measured by real-time RT-PCR in the liver and esophagus of an EGFR transgenic mouse and control wild-type mouse. Mean Ϯ SE (n ϭ 3) in a representative experiment is shown. D. Esophageal epithelial cells (mouse esophageal keratinocytes) isolated from the EGFR transgenic mouse (MEK-EGFR) and the control mouse (MEK-WT) were grown in culture and subjected to real-time RT-PCR for IGFBP-3 mRNA. Mean Ϯ SE (n ϭ 3) in a representative experiment is shown. E. Western blotting was carried out to determine EGFR and IGFBP-3 protein expressed in cell lysates and CM, respectively. Tubulin was used as a loading control. *, P Ͻ 0.01. 7715 IGFBP-3 OVEREXPRESSION IN ESOPHAGEAL CANCER

Fig. 3. IGFBP-3 is overexpressed in primary esophageal cancer tissues. IGFBP-3 mRNA was determined by real-time RT-PCR in paired tissues of tumor and adjacent normal mucosa. Nineteen ESCC (A) and seven EAC (B) samples were analyzed. Mean Ϯ SE (n ϭ 3) in a representative experiment is shown. IGFBP-3 overexpression was based on Ն3-fold IGFBP-3 mRNA levels in tumor specimens (f) compared with normal tissue specimens (Ⅺ). C. EGFR and IGFBP-3 were determined in tissue lysates prepared from 11 paired clinical samples by Western blotting. T, tumor; N, adjacent normal tissue. Tubulin was used as a loading control. esophageal cancer cell lines expressed 30-fold or more IGFBP-3 containing reduced serum concentration (0.5% FCS), it is possible mRNA than parental EPC2 cells at 3 PDs (Fig. 5A). It should be noted that such starving conditions affected IGFBP-3 production in the cells. that presenescent EPC2 cells (42 PDs) expressed 10-fold more IG- In fact, ELISA showed that IGFBP-3 expression level is significantly FBP-3 mRNA than EPC2 cells at early passages (3 PDs), consistent lower in starving medium than in full medium (Fig. 5D). with observations by others about IGFBP-3 expression in senescent To determine whether IGFBP-3 produced by cell lines is functional, cells (10–12). IGFBP-3 mRNA was also increased in EPC1 (data not we performed Western ligand blotting. It is known that only full- shown). When examining nonesophageal cancer cell lines, A431 and length IGFBP-3, but not proteolytically cleaved versions, is capable of HaCaT cells expressed IGFBP-3 mRNA, whereas other cell lines, high-affinity binding to IGF-I (22, 23). Fig. 6A demonstrates that CM including HeLa, HepG2, PANC-1, and MCF7, did not (Fig. 5A; data from A431 and TE2 cells contains whose molecular masses not shown). Western blotting also demonstrated that 6 of 15 (40%) are consistent with IGFBP-3 (40–44 kDa) and IGFBP-4 (24 kDa; ref. esophageal cancer cell lines expressed higher levels of IGFBP-3 than 21), which were documented by reprobing the ligand blot with spe- EPC2 cells (Fig. 5B). cific antibodies (Fig. 6B and C), thus indicating that IGFBP-3 secreted It is possible that rapid secretion of IGFBP-3 into the cell culture by these cancer cell lines can bind to IGF-I. medium after protein synthesis may explain the differences in RNA EGFR Regulates IGFBP-3 Expression and Secretion. Because and protein overexpression. Therefore, we examined the CM for the ELISA data suggested that serum components may be responsible IGFBP-3 by Western blotting and ELISA. Western blotting showed for IGFBP-3 secretion in most of the esophageal cancer cell lines and that IGFBP-3 was detectable in concentrated CM (Fig. 5C), and A431 cells with IGFBP-3 up-regulation (Fig. 5D), we next asked IGFBP-3 protein correlated between cell lysates and the CM (Fig. 5B whether EGFR ligands in serum and corresponding EGFR activation and C). Smaller molecular masses (31–24 kDa) were detected in are involved in regulation of IGFBP-3 expression in these cell lines. addition to the full-length form (Fig. 5C). The smaller molecular As shown in Fig. 7A, AG1478, an inhibitor of EGFR tyrosine kinase masses are compatible with the known proteolytically cleaved forms activity, partially inhibited IGFBP-3 secretion into CM of TE11 in a of IGFBP-3. An ELISA also confirmed that IGFBP-3 protein is highly dose-dependent fashion but did not inhibit IGFBP-3 secretion into expressed in CM (Fig. 5D). Consistent with the Western blotting data, CM of TE2 and TE7 cells. Such an inhibitory effect of AG1478 on IGFBP-3 appeared to be overexpressed in A431 and 6 of 15 (40%) IGFBP-3 expression was further confirmed by ELISA and Western esophageal cancer cell lines compared with primary and immortalized blotting in cell lysates as well as CM from A431 cells (data not human esophageal cells (Fig. 1D). Importantly, anti–IGFBP-3 anti- shown), thereby indicating that serum in the medium can activate bodies used in Western blotting and ELISA did not cross-react with EGFR and lead to IGFBP-3 production in a subset of the cell lines. bovine IGFBP-3 existing in FCS used as a negative control (data not Furthermore, EGF induced IGFBP-3 expression in A431 cells under shown). Because CM for Western blotting was collected with DMEM serum-deprived conditions (0.5% FCS), whereas AG1478 antago- 7716 IGFBP-3 OVEREXPRESSION IN ESOPHAGEAL CANCER

Fig. 4. IGFBP-3 protein is highly expressed in primary esophageal cancer. IGFBP-3 immunohis- tochemistry is shown in tissue sections of (A) nor- mal esophageal epithelium, (B) squamous dyspla- sia, (CϪE) invasive ESCC, and (FϪH) EAC. AϪD represent clinical specimen ESCC4; F and G represent EAC1 in which IGFBP-3 mRNA levels were determined, and H represents EAC6 in which IGFBP-3 mRNA levels were determined (Fig. 3). Intensity was ϩ in A, ϩϩ in B and F, and ϩϩϩ in CϪE, G, and H. The insets in C and G were shown with higher magnification as D and H, re- spectively. Magnification: C, E, and G, ϫ100; A, B, and F, ϫ200; D and H, ϫ400.

nized this effect in a dose-dependent fashion (data not shown). EGF- TE8 cells based on Western blotting (data not shown). It should be mediated induction of IGFBP-3 was also observed in TE11 cells noted that Western blotting (data not shown) confirmed the effects of (Fig. 7B). Thus, these experiments suggest that EGFR activation leads EGF on EGFR tyrosine phosphorylation in these cell lines. to IGFBP-3 induction in A431 and TE11 cells. By contrast, EGF We further asked whether EGFR regulates the level of IGFBP-3 suppressed the basal level of IGFBP-3 expression in TE2 and TE7 secreted in the CM. ELISA demonstrated that a concentration of 1 to cells in a dose-dependent manner when cells were serum starved 100 ng/mL EGF stimulates secretion of IGFBP-3 in TE11 cells overnight (Fig. 7D and F). Interestingly, the IGFBP-3 level was (Fig. 7C) and A431 cells (data not shown). By contrast, 0.1 to 100 further increased in TE2 and TE7 cells when the cells were left serum ng/mL EGF dose-dependently suppressed IGFBP-3 secretion in CM starved for an additional 48 hours (Fig. 7D and F, Lanes 2). EGF also of TE2 and TE7 cells (Fig. 7E and G), whereas 0.1 to 10 pg/mL of inhibited IGFBP-3 expression in TE12 cells in a dose-dependent EGF stimulated IGFBP-3 secretion (Fig. 7E and G), thus suggesting fashion (data not shown). Moreover, EGF stimulation did not appear a biphasic effect of EGF on IGFBP-3 expression in these cell lines. to affect the basal level of IGFBP-3 expression detectable in TE1 and Such changes in IGFBP-3 levels in CM on EGF stimulation appeared 7717 IGFBP-3 OVEREXPRESSION IN ESOPHAGEAL CANCER

Fig. 5. IGFBP-3 is overexpressed in esophageal cancer cell lines. A. IGFBP-3 mRNA was determined by real-time RT-PCR in primary human esophageal cells EPC1 and EPC2; EAC cell line TE7; ESCC cell lines TE1, TE2, TE3, TE5, TE6, TE8, TE9, TE10, TE11, TE12, TE15, T.T, HCE4, and HCE7; and nonesophageal cell lines. Mean Ϯ SE (n ϭ 3) in a representative experiment is shown. B. IGFBP-3 expression was determined by Western blotting using cell lysates from the indicated cell lines. Tubulin was used as a loading control. C. IGFBP-3 secreted by cell lines into DMEM-0.5% FCS was determined by Western blotting using concentrated (10ϫ) CM. D. ELISA was done using CM without concentration. For protein analysis shown in C and D,1ϫ 106 cells were seeded per 100-mm plate. Medium change was done 48 hours before harvesting CM with either full medium (DMEM-10% FCS) or starving medium (DMEM-0.5% FCS). Protein yield in cell lysates prepared simultaneously was used to adjust the difference in cell number that may affect IGFBP-3 concentration in CM. to be consistent with our findings in cell lysates as demonstrated by Inhibition of IGFBP-3 by Small Interfering RNA in TE11 Cells Western blotting (Fig. 7B, D, and F; data not shown). These obser- Augments Cell Proliferation. To gain insight into the biological vations support the premise that EGFR activation regulates IGFBP-3 role of IGFBP-3 in esophageal cancer cell lines, we stably trans- expression and secretion. duced TE11 cells with siRNA directed against IGFBP-3, resulting

Fig. 6. IGFBP-3 secreted by cancer cell lines is capable of binding to IGF-I. A.10ϫ-concentrated CM was examined for 125I -labeled IGF-I binding activity by Western ligand blotting. A nonspecific band (*), present not only in CM from the indicated cell lines but also in 0.5% FCS, is likely bovine IGFBP. The identical blot was reprobed with antibodies spe- cific for IGFBP-4 (B) and IGFBP-3 (C) to identify the IGF-I binding activities shown in A.

7718 IGFBP-3 OVEREXPRESSION IN ESOPHAGEAL CANCER

Fig. 7. IGFBP-3 expression is regulated by EGFR activation. IGFBP-3 expression in cell lysates and CM was determined by ELISA (A, C, E, and G) and Western blotting (B, D, and F) after treatment with AG1478 (specific EGFR tyrosine kinase inhibitor), dimethyl sulfoxide as a control vehicle, or EGF at the indicated concentrations and for the indicated periods. A. TE2, TE7, and TE11 cells were incubated with or without AG1478 for 72 hours in the presence of 10% FCS, and CM was subjected to ELISA. BϪG. TE11 (B and C), TE2 (D and E), and TE7 (F and G) cells were serum starved with serum-free medium for 16 hours and treated with or without EGF for 48 hours. *, P Ͻ 0.01.

7719 IGFBP-3 OVEREXPRESSION IN ESOPHAGEAL CANCER in creation of IGFBP-3 siRNA-expressing TE11-S cells and empty talized human esophageal cells transduced with EGFR and that vector-transduced TE11-N control cells. We confirmed by Western IGFBP-3 is frequently overexpressed in esophageal cancer cell lines blotting that IGFBP-3 protein expression was suppressed by 70% and primary esophageal cancer tissues. A subset of esophageal cancer in TE11-S cells compared with TE11-N cells (Fig. 8A). ELISA also cell lines expresses high levels of both IGFBP-3 and EGFR. revealed a 78% reduction in IGFBP-3 secretion in CM of TE11-S What is the basis for EGFR-mediated regulation of IGFBP-3? One cells (Fig. 8B). When cell growth was assessed, TE11-S cells possibility is that IGFBP-3 expression is subject to regulation by appeared to grow faster than control cells (Fig. 8C). This notion EGFR signaling. Additionally, EGFR overexpression in EPC2 3 was documented by a [ H]thymidine incorporation assay showing cells induces a phenotypic change associated with growth inhibition, 14-fold more active DNA synthesis in TE11-S cells than in earlier senescence, or differentiation, culminating in up-regulation of TE11-N cells (Fig. 8D). Moreover, DNA content analysis by flow IGFBP-3. Akerman et al. (24) observed that excessive EGFR signal- cytometry revealed a decrease in G1 phase (10%) and an increase ing shortened the replicative life span of normal human skin kerati- in S phase (10%) in TE11-S cells compared with TE11-N cells nocytes. IGFBP-3 has been found to be up-regulated in senescent or (data not shown). No sub-G fraction was observed in both cell 1 quiescent human diploid fibroblasts (11), vascular endothelial cells lines when they were grown under standard tissue culture condi- (12), and prostate epithelial cells (14). Thus, IGFBP-3 induction in tions or in a serum-starved condition for 48 hours (data not shown). EPC2-EGFR and EPC2-hTERT-EGFR cells may be associated with The trypan blue dye exclusion test also detected no significant growth inhibition in monolayer culture. By contrast, it is unlikely that difference in cell viability rate between TE11-S and TE11-N cells. IGFBP-3 up-regulation is a consequence of EPC2-EGFR cell differ- These observations indicate that IGFBP-3 may restrain TE11 cell proliferation without inducing . However, we would em- entiation because our gene array data did not show any changes for phasize that further investigation is needed. early and terminal keratinocyte differentiation markers, including cytokeratins K1/K10, K4/K13, involucrin, and profilaggrin (data not shown). In addition, EPC2-EGFR cells are capable of undergoing DISCUSSION differentiation in organotypic culture (6). By contrast, IGFBP-3 up- An unbiased gene array-based screening allowed us to identify regulation is associated with esophageal epithelial cell hyperprolifera- IGFBP-3 as a novel gene critically regulated in EPC2 cells that tion induced by EGFR overexpression in both organotypic cell culture overexpress EGFR. This was further confirmed in independently and transgenic mice (Fig. 2). Thus, regulation of IGFBP-3 by EGFR transduced primary and hTERT-immortalized esophageal epithelial in a physiologic context may be different from that observed in cells. We found that IGFBP-3 is up-regulated in primary and immor- monolayer cell culture.

Fig. 8. Cell growth is enhanced in TE11 cells by siRNA-mediated suppression of IGFBP-3. Expression of IGFBP-3 in TE11 cells stably transduced with siRNA against IGFBP-3 (TE11-S) and control cells (TE11-N) was determined by Western blotting (A) and ELISA (B). A total of 1 ϫ 106 cells were seeded per 100-mm plate, and medium was exchanged 48 hours before harvesting cell lysates and CM. ␤-Actin was used as a loading control for Western blotting. Protein yields in cell lysates were used to adjust the differences in cell number that may affect IGFBP-3 concentration in CM. C. Growth curves were drawn by seeding cells in 6-well plates (5.0 ϫ 104 cells per well) in a triplicate fashion and counting them at the indicated time points. D. DNA synthesis was measured in subconfluent TE11-S and TE11-N cells. Mean Ϯ SE (n ϭ 4) represents one of two independently performed experiments with similar results. *, P Ͻ 0.01. 7720 IGFBP-3 OVEREXPRESSION IN ESOPHAGEAL CANCER

IGFBP-3 is transcriptionally activated by p53 (25, 26) through taneously immortalized human skin keratinocytes (33, 34). These several p53-binding cis-regulatory elements (25, 27). Given the role observations are consistent with our observations in TE2 and TE7 of p53 in cellular senescence (28), it is intriguing to speculate that p53 cells when they were stimulated at nanogram levels of EGF transcriptional activity is involved in IGFBP-3 induction in EPC2- (Fig. 7DϪG). By contrast, picogram levels of EGF induced IGFBP-3 EGFR cells. In fact, we have recently observed that p53R175H inhib- in these cell lines, albeit modestly (Fig. 7E and G). Furthermore, EGF ited IGFBP-3 mRNA up-regulation and prolonged the replicative life potently induced IGFBP-3 expression and secretion in TE11 and span of EPC2-EGFR cells (data not shown). A431 cells (Fig. 7B and C; data not shown). EGF-mediated induction Mechanisms for IGFBP-3 Overexpression in Esophageal Can- of IGFBP-3 secretion has been demonstrated in fetal rat lung fibro- cer. Our finding of IGFBP-3 overexpression in esophageal cancer is blasts (35) and newborn rat astroblasts (36). Such a discrepancy in the first such demonstration in this cancer type. Our data showed a EGF response among cell lines may be accounted for by differences significant correlation between mRNA levels and protein levels in in EGFR level or activation. Additionally, we also found that higher cancer cell lines as well. It is noteworthy that EGFR and IGFBP-3 are cell density enhances EGF-mediated induction of IGFBP-3 expression localized in close proximity on human chromosome 7p12.3-p12.1 and in these esophageal cancer cell lines (data not shown). 7p13-p12, respectively. Because EGFR gene amplification is noted in Feldser et al. (37) recently showed that IGFBP-3 expression is several esophageal cancer cell lines including TE3, TE5, and TE8 (29) prominently impaired in hypoxia-inducible factor (HIF)-1␣Ϫ/Ϫ em- and A431 cells, it is interesting to consider the possibility of coam- bryonic stem cells and that hypoxia induces IGFBP-3 mRNA in plification of the IGFBP-3 and EGFR loci. However, gene amplifi- HIF-1␣ϩ/ϩ embryonic stem cells. They also showed that hypoxia, as cation seems to be the least likely mechanism for IGFBP-3 overex- well as fibroblast growth factor and EGF, induces HIF-1␣ in 293 pression in the majority of cases because EGFR gene amplification is cells. Functional inactivation of the von Hippel-Lindau (VHL) gene, found in only 10% to 20% of primary esophageal cancer tissues that an essential component of E3 ubiquitin-ligase complex, is common in overexpress EGFR (5, 30). Hintz et al. (31) also found no chromo- renal cancer and known to result in HIF-1␣ stabilization. However, somal alterations such as recombination and gene amplification in VHL deficiency is not common in esophageal cancer (38–40). Treins renal tumors that overexpress IGFBP-3. et al. (41) showed that insulin activates HIF-1␣ through the phos- ELISA and Western blotting showed remarkable down-regulation phatidylinositol 3Ј-kinase/AKT pathway in retinal epithelial cells. of IGFBP-3 in most of the cell lines examined on serum deprivation Because EGFR activates AKT in esophageal cancer cell lines (19) and (Fig. 5D), thus suggesting that serum factors play a role in regulation EPC2-EGFR cells (6), it is tempting to speculate that HIF-1␣ may of IGFBP-3 expression in these cells. Serum deprivation could affect play a role in IGFBP-3 induction in esophageal cancer cells as well as cell growth and thus production of IGFBP-3. An equivalent number of in primary and immortalized human esophageal cells transduced with cells were seeded per plate to collect conditioned media for ELISA. EGFR. However, differences in cell number on harvesting of conditioned Growth hormone, a potent inducer of IGFBP-3, activates Jak2 to media could affect the IGFBP-3 concentrations. Thus, we calibrated induce tyrosine phosphorylation of EGFR, independent of its own the ELISA data on conditioned media using total protein yield from ligands and intrinsic kinase activity, resulting in recruitment of Grb2 simultaneously harvested cell lysates. and activation of -activated protein kinases (42). Thus, it is In this study, EGFR tyrosine kinase activity was involved in the also possible that growth hormone utilizes EGFR to induce IGFBP-3 regulation of IGFBP-3 expression in several esophageal cancer cell in the absence of EGFR ligands. Interestingly, a prospective study on lines and A431 cells. A431 and TE11 cells exhibited partial depend- a cohort of 1,041 men with acromegaly revealed an increased inci- ency of IGFBP-3 expression on EGFR tyrosine kinase activity when dence rate for esophageal cancer (43). they were grown in the presence of 10% FCS (Fig. 7A; data not Our data also clearly suggest that p53 is unlikely to contribute to shown). By contrast, suppression of EGFR kinase activity did not IGFBP-3 induction in cancer cell lines, contrary to primary human influence the IGFBP-3 level in TE2 and TE7 cells (Fig. 7A), suggest- esophageal cells, because p53 mutation and/or loss of p53 functions ing that EGFR ligands in serum may not be the only factors contrib- have been documented in almost all esophageal cancer cell lines uting to IGFBP-3 induction. In addition, the IGFBP-3 level maximally (44, 45). induced in CM by EGF in TE11 and A431 cells appeared to be lower The Role of IGFBP-3 in Regulation of Esophageal Cell Prolif- than that induced by 10% FCS (Figs. 5D and 7C; data not shown). In eration and Apoptosis. IGFBP-3 has been shown to both stimulate primary esophageal tumor tissues, IGFBP-3 expression level tended to and inhibit cell proliferation under various experimental conditions correlate with EGFR level (Fig. 3C). However, three samples through IGF-dependent and independent mechanisms (9, 13). How- (ESCC9, ESCC13, and ESCC20) appeared to express an increased ever, IGFBP-3 plays a critical role in the induction of apoptosis of level of IGFBP-3 without EGFR overexpression (Fig. 3C), suggesting several cancer cells in a p53-dependent (46) and -independent (47, 48) a separate mechanism accounting for the IGFBP-3 induction in these fashion. In addition, many agents have been shown to inhibit breast samples. cancer cell proliferation in vitro through induction and secretion of Numerous agents other than EGF have been shown to induce IGFBP-3 (48). They include tumor necrosis factor ␣, transforming IGFBP-3 in vitro (9, 13). They include growth hormone, IGF-I, growth factor ␤, retinoic acid, vitamin D, and antiestrogen compounds transforming growth factor ␤, tumor necrosis factor ␣, retinoic acid, such as tamoxifen and ICI182,780. Recently, silibinin, a flavonoid and vitamin D. Thus, such peptide growth factors and hormones may antioxidant derived from milk thistle, has been shown to up-regulate contribute to IGFBP-3 induction in cell lines and primary tumor IGFBP-3 to inhibit proliferation of prostate cancer cells (49). Inter- tissues. They may also affect cellular responses to EGFR ligands in estingly, EGFR appeared to mediate the cytotoxic effect of silibinin modulating IGFBP-3 expression. For example, all-trans-retinoic acid on rat glioma cells (50). induces IGFBP-3 in the presence of EGF at concentrations that Thus, the antiproliferative or proapoptotic activities of IGFBP-3 suppress IGFBP-3 in cervical epithelial cell lines (32). Further com- may provide a safeguard mechanism against carcinogenesis in certain plexity underlying EGFR-mediated regulation of IGFBP-3 was ob- tumor types. In fact, down-regulation of IGFBP-3 has been docu- served in that the effect of EGF stimulation on IGFBP-3 expression is mented in several cancers, including prostate cancer (51) and hepa- different among the cell lines examined. EGF has been shown to tocellular carcinoma (52, 53). Consistent with these observations, inhibit IGFBP-3 mRNA and protein expression in primary and spon- IGFBP-3 was down-regulated in prostate epithelial cells immortalized 7721 IGFBP-3 OVEREXPRESSION IN ESOPHAGEAL CANCER by human papilloma virus E7 oncoprotein through a mechanism REFERENCES involving physical interaction between E7 and IGFBP-3, leading to 1. Nakagawa H, Katzka D, Rustgi AK. Biology of esophageal cancer. In: Rustgi AK, polyubiquitination and proteolytic degradation of IGFBP-3 (14, 54), editor. Gastrointestinal cancers. London: Elsevier; 2003. p. 241–51. although another study (55) showed up-regulation of IGFBP-3 in 2. Schlessinger J. Ligand-induced, receptor-mediated dimerization and activation of EGF receptor. Cell 2002;110:669–72. cervical keratinocytes immortalized with human papillomavirus E6 3. Velu TJ, Beguinot L, Vass WC, et al. Epidermal-growth-factor-dependent transfor- and E7. Hanafusa et al. (56) showed epigenetic silencing of IGFBP-3 mation by a human EGF receptor proto-oncogene. Science (Wash DC) 1987;238: 1408–10. transcription through hypermethylation of the IGFBP-3 promoter as a 4. Yacoub L, Goldman H, Odze RD. Transforming growth factor-alpha, epidermal mechanism for IGFBP-3 down-regulation in hepatocellular carci- growth factor receptor, and MiB-1 expression in Barrett’s-associated neoplasia: noma. correlation with prognosis. Mod Pathol 1997;10:105–12. 5. Itakura Y, Sasano H, Shiga C, et al. Epidermal growth factor receptor overexpression Our findings of IGFBP-3 overexpression in esophageal cancer in esophageal carcinoma. An immunohistochemical study correlated with clinico- represent a sharp contrast to these previous observations. One inter- pathologic findings and DNA amplification. Cancer (Phila) 1994;74:795–804. 6. Andl CD, Mizushima T, Nakagawa H, et al. Epidermal growth factor receptor pretation is that IGFBP-3 may stimulate cell proliferation by enriching mediates increased cell proliferation, migration, and aggregation in esophageal kera- IGF on the cell surface and thereby allowing it to activate the IGF tinocytes in vitro and in vivo. J Biol Chem 2003;278:1824–30. receptor, following the proteolytic cleavage of IGFBP-3 mediated by 7. Harada H, Nakagawa H, Oyama K, et al. Telomerase induces immortalization of human esophageal keratinocytes without p16INK4a inactivation. Mol Cancer Res enzymes such as matrix metalloproteases (9, 57). Another possibility 2003;1:729–38. is that IGFBP-3 was induced in esophageal cancer cells as a result of 8. Hwa V, Oh Y, Rosenfeld RG. The insulin-like growth factor-binding protein (IGFBP) superfamily. Endocr Rev 1999;20:761–87. exposure to various growth factors, hormones, and cellular stress such 9. Firth SM, Baxter RC. Cellular actions of the insulin-like growth factor binding as hypoxia and therapeutic agents in tumor tissues. However, a subset proteins. Endocr Rev 2002;23:824–54. of these cancer cells acquires genetic alterations to negate the growth- 10. Wraight CJ, Murashita MM, Russo VC, Werther GA. A keratinocyte cell line synthesizes a predominant insulin-like growth factor-binding protein (IGFBP-3) that inhibitory effect of IGFBP-3. Consistent with such a notion is the modulates insulin-like growth factor-I action. J Investig Dermatol 1994;103:627–31. finding that growth inhibition on excessive EGFR activation observed 11. Goldstein S, Moerman EJ, Jones RA, Baxter RC. Insulin-like growth factor binding protein 3 accumulates to high levels in culture medium of senescent and quiescent in A431 cells is not found in the majority of esophageal cancer cell human fibroblasts. Proc Natl Acad Sci USA 1991;88:9680–4. lines (58). Another possibility is that IGFBP-3 inhibits tumor cell 12. Grillari J, Hohenwarter O, Grabherr RM, Katinger H. Subtractive hybridization of proliferation as a negative feedback mechanism. EGF treatment of mRNA from early passage and senescent endothelial cells. Exp Gerontol 2000;35: 187–97. human papilloma virus type 16-immortalized ectocervical epithelial 13. Grimberg A, Cohen P. Role of insulin-like growth factors and their binding proteins cell lines results in decreased IGFBP-3 expression and concomitant in growth control and carcinogenesis. J Cell Physiol 2000;183:1–9. 14. Schwarze SR, DePrimo SE, Grabert LM, et al. Novel pathways associated with enhancement of cell proliferation (32, 59). Thus, it is possible that bypassing cellular senescence in human prostate epithelial cells. J Biol Chem 2002; EGF-induced down-regulation of IGFBP-3 in TE2 and TE7 cells may 277:14877–83. facilitate the mitogenic effect of EGF in these cell lines. 15. Modric T, Silha JV, Shi Z, et al. Phenotypic manifestations of insulin-like growth factor-binding protein-3 overexpression in transgenic mice. Endocrinology 2001;142: It is noteworthy that IGFBP-3 up-regulation was observed in esoph- 1958–67. ageal squamous dysplasia by immunohistochemistry. It should be 16. Murphy LJ, Molnar P, Lu X, Huang H. Expression of human insulin-like growth factor-binding protein-3 in transgenic mice. J Mol Endocrinol 1995;15:293–303. emphasized that IGFBP-3 has been demonstrated to enhance EGFR 17. Neuenschwander S, Schwartz A, Wood TL, et al. Involution of the lactating mam- phosphorylation, activation of p44/42 and p38 mitogen-activated pro- mary gland is inhibited by the IGF system in a transgenic mouse model. J Clin tein kinases, and cell proliferation in MCF10A breast epithelial cells Investig 1996;97:2225–32. 18. Nakagawa H, Wang TC, Zukerberg L, et al. The targeting of the cyclin D1 oncogene stimulated with 0.1 to 1 ng/mL EGF (60). Because EGFR overex- by an Epstein-Barr virus promoter in transgenic mice causes dysplasia in the tongue, pression is one of the earliest genetic changes found in esophageal esophagus and forestomach. Oncogene 1997;14:1185–90. 19. Okano J, Gaslightwala I, Birnbaum MJ, Rustgi AK, Nakagawa H. Akt/protein kinase carcinogenesis, it is intriguing to speculate that the interplay of EGFR B isoforms are differentially regulated by epidermal growth factor stimulation. J Biol and IGFBP-3 contributes to aberrant cell growth of the esophageal Chem 2000;275:30934–42. epithelium. 20. Opitz OG, Harada H, Suliman Y, et al. A mouse model of human oral-esophageal cancer. J Clin Investig 2002;110:761–9. In aggregate, we have identified IGFBP-3 as an aberrantly regu- 21. Coleman CM, Grimberg A. Alterations in the expression of insulin-like growth lated gene through the EGFR signaling pathway. IGFBP-3 overex- factors and their binding proteins in lung cancer. Methods Mol Med 2003;74:167–86. 22. Lalou C, Lassarre C, Binoux M. A proteolytic fragment of insulin-like growth factor pression in esophageal cancer implies a role in esophageal carcino- (IGF) binding protein-3 that fails to bind IGFs inhibits the mitogenic effects of IGF-I genesis through IGF-dependent or independent actions. Additional and insulin. Endocrinology 1996;137:3206–12. studies are required to elucidate how IGFBP-3 affects cellular func- 23. Lalou C, Lassarre C, Binoux M. Isolation and characterization of proteolytic frag- ments of insulin-like growth factor-binding protein-3. Horm Res 1996;45:156–9. tions, such as proliferation, differentiation, and migration, and the 24. Akerman GS, Tolleson WH, Brown KL, et al. Human papillomavirus type 16 E6 and interplay between EGFR and IGFBP-3. E7 cooperate to increase epidermal growth factor receptor (EGFR) mRNA levels, overcoming mechanisms by which excessive EGFR signaling shortens the life span of normal human keratinocytes. Cancer Res 2001;61:3837–43. 25. Buckbinder L, Talbott R, Velasco-Miguel S, et al. Induction of the growth inhibitor ACKNOWLEDGMENTS IGF-binding protein 3 by p53. Nature (Lond) 1995;377:646–9. 26. Grimberg A. P53 and IGFBP-3: apoptosis and cancer protection. Mol Genet Metab 2000;70:85–98. We thank Dr. Yasuhiro Shirakawa (Okayama University, Okayama, Japan) 27. Bourdon JC, Deguin-Chambon V, Lelong JC, et al. Further characterisation of the p53 for clinical samples. We thank Dr. Don Baldwin (Microarray facility at the responsive element: identification of new candidate genes for trans-activation by p53. University of Pennsylvania, School of Medicine) for gene array experiments, Oncogene 1997;14:85–94. Dr. Gary Swain (Morphology Core facility) for immunohistochemistry, Dr. 28. Donehower LA. Does p53 affect organismal aging? J Cell Physiol 2002;192:23–33. 29. Yamamoto T, Kamata N, Kawano H, et al. High incidence of amplification of the Sue A. Keilbaugh and Weimian He (Molecular Biology Core facility) for epidermal growth factor receptor gene in human squamous carcinoma cell lines. technical assistance in RNA isolation and real-time RT-PCR, Caitlin Smith and Cancer Res 1986;46:414–6. Michael K. Mashiba for technical assistance in ELISA and Western blotting, 30. Kitagawa Y, Ueda M, Ando N, et al. Further evidence for prognostic significance of epidermal growth factor receptor gene amplification in patients with esophageal and Drs. Therese B. Deramaudt, Cameron N. Johnstone, and Ben Rhoades squamous cell carcinoma. Clin Cancer Res 1996;2:909–14. (Rustgi laboratory) for helpful discussions at the University of Pennsylvania, 31. Hintz RL, Bock S, Thorsson AV, et al. Expression of the insulin like growth Philadelphia, PA. Finally, we are grateful to Drs. Meenhard M. Herlyn (Wistar factor-binding protein 3 (IGFBP-3) gene is increased in human renal carcinomas. Institute, Philadelphia, PA) and Anil K. Rustgi (Gastroenterology Division, J Urol 1991;146:1160–3. 32. Andreatta-Van Leyen S, Hembree JR, Eckert RL. Regulation of insulin-like growth University of Pennsylvania) for their support and advice and for critical review factor 1 binding protein 3 levels by epidermal growth factor and retinoic acid in of the manuscript by Dr. Rustgi. cervical epithelial cells. J Cell Physiol 1994;160:265–74. 7722 IGFBP-3 OVEREXPRESSION IN ESOPHAGEAL CANCER

33. Wraight CJ, Werther GA. Insulin-like growth factor-I and epidermal growth factor 48. Baxter RC. Signalling pathways involved in antiproliferative effects of IGFBP-3: a regulate insulin-like growth factor binding protein-3 (IGFBP-3) in the human kera- review. Mol Pathol 2001;54:145–8. tinocyte cell line HaCaT. J Investig Dermatol 1995;105:602–7. 49. Zi X, Zhang J, Agarwal R, Pollak M. Silibinin up-regulates insulin-like growth 34. Edmondson SR, Murashita MM, Russo VC, Wraight CJ, Werther GA. Expression of factor-binding protein 3 expression and inhibits proliferation of androgen-indepen- insulin-like growth factor binding protein-3 (IGFBP-3) in human keratinocytes is dent prostate cancer cells. Cancer Res 2000;60:5617–20. regulated by EGF and TGFbeta1. J Cell Physiol 1999;179:201–7. 50. Qi L, Singh RP, Lu Y, et al. Epidermal growth factor receptor mediates silibinin- 35. Price WA. Peptide growth factors regulate insulin-like growth factor binding protein induced cytotoxicity in a rat glioma cell line. Cancer Biol Ther 2003;2:526–31. production by fetal rat lung fibroblasts. Am J Respir Cell Mol Biol 1999;20:332–41. 51. Tennant MK, Thrasher JB, Twomey PA, Birnbaum RS, Plymate SR. Insulin-like 36. Loret C, Janet T, Labourdette G, Schneid H, Binoux M. FGFs stimulate IGF binding growth factor-binding protein-2 and -3 expression in benign human prostate epithe- protein synthesis without affecting IGF synthesis in rat astroblasts in primary culture. lium, prostate intraepithelial neoplasia, and adenocarcinoma of the prostate. J Clin Glia 1991;4:378–83. Endocrinol Metab 1996;81:411–20. 37. Feldser D, Agani F, Iyer NV, et al. Reciprocal positive regulation of hypoxia- 52. Gong Y, Cui L, Minuk GY. The expression of insulin-like growth factor binding inducible factor 1alpha and insulin-like growth factor 2. Cancer Res 1999;59:3915–8. proteins in human hepatocellular carcinoma. Mol Cell Biochem 2000;207:101–4. 38. Kuroki T, Trapasso F, Yendamuri S, et al. Allele loss and promoter hypermethylation 53. Huynh H, Chow PK, Ooi LL, Soo KC. A possible role for insulin-like growth of VHL, RAR-beta, RASSF1A, and FHIT tumor suppressor genes on chromosome 3p factor-binding protein-3 autocrine/paracrine loops in controlling hepatocellular car- in esophageal squamous cell carcinoma. Cancer Res 2003;63:3724–8. cinoma cell proliferation. Cell Growth Differ 2002;13:115–22. 39. Nie Y, Liao J, Zhao X, et al. Detection of multiple gene hypermethylation in the 54. Mannhardt B, Weinzimer SA, Wagner M, et al. Human papillomavirus type 16 E7 development of esophageal squamous cell carcinoma. Carcinogenesis (Lond) 2002; oncoprotein binds and inactivates growth-inhibitory insulin-like growth factor bind- 23:1713–20. ing protein 3. Mol Cell Biol 2000;20:6483–95. 40. Wijnhoven BP, Lindstedt EW, Abbou M, et al. Molecular genetic analysis of the von 55. Berger AJ, Baege A, Guillemette T, et al. Insulin-like growth factor-binding protein Hippel-Lindau and human peroxisome proliferator-activated receptor gamma tumor- 3 expression increases during immortalization of cervical keratinocytes by human suppressor genes in adenocarcinomas of the gastroesophageal junction. Int J Cancer 2001;94:891–5. papillomavirus type 16 E6 and E7 proteins. Am J Pathol 2002;161:603–10. 41. Treins C, Giorgetti-Peraldi S, Murdaca J, Semenza GL, Van Obberghen E. Insulin 56. Hanafusa T, Yumoto Y, Nouso K, et al. Reduced expression of insulin-like growth stimulates hypoxia-inducible factor 1 through a phosphatidylinositol 3-kinase/target factor binding protein-3 and its promoter hypermethylation in human hepatocellular of rapamycin-dependent signaling pathway. J Biol Chem 2002;277:27975–81. carcinoma. Cancer Lett 2002;176:149–58. 42. Yamauchi T, Ueki K, Tobe K, et al. Tyrosine phosphorylation of the EGF receptor by 57. Fowlkes JL, Enghild JJ, Suzuki K, Nagase H. Matrix metalloproteinases degrade the kinase Jak2 is induced by growth hormone. Nature (Lond) 1997;390:91–6. insulin-like growth factor-binding protein-3 in dermal fibroblast cultures. J Biol 43. Ron E, Gridley G, Hrubec Z, et al. Acromegaly and gastrointestinal cancer. Cancer Chem 1994;269:25742–6. (Phila) 1991;68:1673–7. 58. Watanabe G, Kaganoi J, Imamura M, et al. Progression of esophageal carcinoma by 44. Jia LQ, Osada M, Ishioka C, et al. Screening the p53 status of human cell lines using loss of EGF-STAT1 pathway. Cancer J 2001;7:132–9. a yeast functional assay. Mol Carcinog 1997;19:243–53. 59. Hembree JR, Agarwal C, Eckert RL. Epidermal growth factor suppresses insulin-like 45. Barnas C, Martel-Planche G, Furukawa Y, et al. Inactivation of the p53 protein in cell growth factor binding protein 3 levels in human papillomavirus type 16-immortalized lines derived from human esophageal cancers. Int J Cancer 1997;71:79–87. cervical epithelial cells and thereby potentiates the effects of insulin-like growth 46. Grimberg A, Liu B, Bannerman P, El-Deiry WS, Cohen P. IGFBP-3 mediates factor 1. Cancer Res 1994;54:3160–6. p53-induced apoptosis during serum starvation. Int J Oncol 2002;21:327–35. 60. Martin JL, Weenink SM, Baxter RC. Insulin-like growth factor-binding protein-3 47. Butt AJ, Firth SM, King MA, Baxter RC. Insulin-like growth factor-binding protein-3 potentiates epidermal growth factor action in MCF-10A mammary epithelial cells. modulates expression of Bax and Bcl-2 and potentiates p53-independent radiation- Involvement of p44/42 and p38 mitogen-activated protein kinases. J Biol Chem induced apoptosis in human breast cancer cells. J Biol Chem 2000;275:39174–81. 2003;278:2969–76.

7723