Chromosomal abnormalities and novel disease-related regions in progression from Barrett's esophagus to esophageal adenocarcinoma

Authors

  • Tadayuki Akagi,

    Corresponding author
    1. Division of Hematology and Oncology, Cedars-Sinai Medical Center, UCLA School of Medicine, Los Angeles, CA
    Current affiliation:
    1. Department of Stem Cell Biology, Graduate School of Medical Science, Kanazawa University, 13-1 Takara-machi, Kanazawa, Ishikawa 920-8640, Japan
    • Division of Hematology and Oncology, Cedars-Sinai Medical Center, UCLA School of Medicine, 8700 Beverly Blvd, Los Angeles, CA 90048, USA
    Search for more papers by this author
    • Fax: +1-310-423-0225.

  • Tetsuo Ito,

    1. Division of Gastroenterology, Department of Medicine, Johns Hopkins School of Medicine, Baltimore, MD
    Search for more papers by this author
  • Motohiro Kato,

    1. Department of Hematology and Oncology, Graduate School of Medicine, University of Tokyo, Tokyo, Japan
    2. Department of Cell Therapy and Transplantation Medicine and the 21st Century COE Program, Graduate School of Medicine, University of Tokyo, Tokyo, Japan
    Search for more papers by this author
  • Zhe Jin,

    1. Division of Gastroenterology, Department of Medicine, Johns Hopkins School of Medicine, Baltimore, MD
    Search for more papers by this author
  • Yulan Cheng,

    1. Division of Gastroenterology, Department of Medicine, Johns Hopkins School of Medicine, Baltimore, MD
    Search for more papers by this author
  • Takatsugu Kan,

    1. Division of Gastroenterology, Department of Medicine, Johns Hopkins School of Medicine, Baltimore, MD
    Search for more papers by this author
  • Go Yamamoto,

    1. Department of Hematology and Oncology, Graduate School of Medicine, University of Tokyo, Tokyo, Japan
    2. Department of Cell Therapy and Transplantation Medicine and the 21st Century COE Program, Graduate School of Medicine, University of Tokyo, Tokyo, Japan
    Search for more papers by this author
  • Alexandru Olaru,

    1. Division of Gastroenterology, Department of Medicine, Johns Hopkins School of Medicine, Baltimore, MD
    Search for more papers by this author
  • Norihiko Kawamata,

    1. Division of Hematology and Oncology, Cedars-Sinai Medical Center, UCLA School of Medicine, Los Angeles, CA
    Search for more papers by this author
  • Jessica Boult,

    1. Division of Hematology and Oncology, Cedars-Sinai Medical Center, UCLA School of Medicine, Los Angeles, CA
    Search for more papers by this author
  • Harmik J. Soukiasian,

    1. Department of Surgery, Cedars-Sinai Medical Center, Los Angeles, CA
    Search for more papers by this author
  • Carl W. Miller,

    1. Division of Hematology and Oncology, Cedars-Sinai Medical Center, UCLA School of Medicine, Los Angeles, CA
    Search for more papers by this author
  • Seishi Ogawa,

    1. Department of Hematology and Oncology, Graduate School of Medicine, University of Tokyo, Tokyo, Japan
    2. Department of Cell Therapy and Transplantation Medicine and the 21st Century COE Program, Graduate School of Medicine, University of Tokyo, Tokyo, Japan
    Search for more papers by this author
  • Stephen J. Meltzer,

    Corresponding author
    1. Division of Gastroenterology, Department of Medicine, Johns Hopkins School of Medicine, Baltimore, MD
    2. Department of Oncology, Sidney Kimmel Comprehensive Cancer Center, Johns Hopkins School of Medicine, Baltimore, MD
    • Division of Gastroenterology, Department of Medicine, Johns Hopkins University School of Medicine, 1503 E. Jefferson St., Room 112, Baltimore, MD 21287, USA
    Search for more papers by this author
    • Fax: +1-410-502-1329.

    • Stephen J. Meltzer and H. Phillip Koeffler are co-senior authors.

  • H. Phillip Koeffler

    1. Division of Hematology and Oncology, Cedars-Sinai Medical Center, UCLA School of Medicine, Los Angeles, CA
    Search for more papers by this author
    • Stephen J. Meltzer and H. Phillip Koeffler are co-senior authors.


Abstract

Barrett's esophagus (BE) is a metaplastic condition caused by chronic gastroesophageal reflux which represents an early step in the development of esophageal adenocarcinoma (EAC). Single-nucleotide polymorphism microarray (SNP-chip) analysis is a novel, precise, high-throughput approach to examine genomic alterations in neoplasia. Using 250K SNP-chips, we examined the neoplastic progression of BE to EAC, studying 11 matched sample sets: 6 sets of normal esophagus (NE), BE and EAC, 4 of NE and BE and 1 of NE and EAC. Six (60%) of 10 total BE samples and 4 (57%) of 7 total EAC samples exhibited 1 or more genomic abnormalities comprising deletions, duplications, amplifications and copy-number-neutral loss of heterozygosity (CNN-LOH). Several shared abnormalities were identified, including chromosome 9p CNN-LOH [2 BE samples (20%)], deletion of CDKN2A [4 BE samples (40%)] and amplification of 17q12-21.2 involving the ERBB2, RARA and TOP2A genes [3.1 Mb, 2 EAC (29%)]. Interestingly, 1 BE sample contained a homozygous deletion spanning 9p22.3–p22.2 (1.2 Mb): this region harbors only 1 known gene, basonuclin 2 (BNC2). Real-time PCR analysis confirmed the deletion of this gene and decreased the expression of BNC2 mRNA in the BE sample. Furthermore, transfection and stable expression of BNC2 caused growth arrest of OE33 EAC cells, suggesting that BNC2 functions as a tumor suppressor gene in the esophagus and that deletion of this gene occurs during the development of EAC. Thus, this SNP-chip analysis has identified several early cytogenetic events and novel candidate cancer-related genes that are potentially involved in the evolution of BE to EAC. © 2009 UICC

Chronic gastroesophageal reflux disease (GERD) is characterized by the retrograde movement of gastric contents into the esophagus, resulting in tissue damage. GERD is the major risk factor for the development of Barrett's esophagus (BE).1 BE is a premalignant condition, greatly increasing the risk of developing esophageal adenocarcinoma (EAC).2, 3

Genomic DNA alterations often contribute to the development of malignant tumors. In BE and EAC, chromosomal aberrations have been discovered by comparative genomic hybridization (CGH) analysis.4–11 CGH analyses have revealed frequent gains of chromosomes 6p (10–37%), 7q (17–37%), 7p (30–60%), 8q (50–80%), 10q (20–50%), 15q (10–40%), 17q (30–50%) and 20q (50–80%); and frequent losses of chromosomes 4q (20–50%), 5q (20–50%), 9p (20–50%), 14q (30–40%), 16q (36–40%), 17p (30%), 18q (20–60%) and Y (60–76%).4–11 These chromosomal alterations have suggested genes associated with esophageal adenocarcinogenesis. For example, the proto-oncogenes MYC (8q), EGFR (7p) and ERBB2 (17q) are often duplicated. The tumor suppressor genes APC, CDKN2A, TP53 and SMAD4 are located on 5q, 9p, 17p and 18q, respectively, and these chromosomal regions are often deleted. Thus, genome-wide analyses of DNA copy-number changes in BE and EAC can identify consensus regions of chromosomal gain and loss, as well as candidate cancer-related genes.

Single-nucleotide polymorphism microarray (SNP-chip) analysis is a novel strategy to examine genomic alterations such as copy-number changes and loss of heterozygosity (LOH).12–14 Importantly, SNP-chip analysis can detect several abnormalities including copy-number-neutral loss of heterozygosity (CNN-LOH) that cannot be detected by either karyotyping or CGH. SNP-chip analysis has been used to study several types of leukemia, including chronic lymphocytic leukemia (CLL),15, 16 childhood acute lymphoblastic leukemia (ALL)17, 18 and acute myeloid leukemia (AML).19–24

In our study, we identified chromosomal abnormalities and novel disease-related genomic regions using 250K SNP-chip analysis in matched tissue sample sets that contained normal esophagus (NE), EAC and/or BE. The use of the CNAG (copy-number analysis for Affymetrix GeneChips) program and the new AsCNAR (allele-specific copy-number analysis using anonymous references) algorithm12, 14 provided a highly sensitive technique to detect CNN-LOH as well as copy-number changes in premalignant and malignant esophageal tissue samples.

Material and methods

Patient samples, isolation of genomic DNA and RNA and cell culture

Samples of BE and EAC tissues were obtained by endoscopic biopsy. Clinical features of esophageal samples examined in this study are summarized in Table I. Normal esophageal mucosal (NE) samples were always obtained from the same individual for BE and EAC samples. Tissue samples were snap-frozen in liquid nitrogen. All NE samples were obtained at a minimum of 7 cm proximal to the squamocolumnar junction (the proximal border between BE and NE). All biopsies were examined histopathologically by hematoxylin and eosin staining; and the stromal component averaged ∼50% and 30% in the BE and EAC samples, respectively. Genomic DNA and total RNA were isolated from tissues using a DNeasy Tissue Kit (Qiagen, Valencia, CA) and Trizol reagent (Invitrogen, Carlsbad, CA), respectively. DNA was determined to be of high-MW by agarose gel electrophoresis showing >90% of DNA to be above a length of 20 kb. The EAC cell line, OE33, was maintained in RPMI 1640 medium (Mediatech, Herndon, VA) supplemented with 10% fetal bovine serum (Atlanta Biological, Lawrenceville, GA).

Table I. Clinical Features of Esophageal Samples Examined by SNP-Chip Analysis
CaseOriginProcedureHistologyAgeGenderRaceEAC gradeTNMStagePrevious treatment
  • Clinical features of 11 samples examined by SNP-chip analysis are displayed.–

  • Abbreviations: NE, normal esophagus; BE, Barrett's esophagus; LGD, low-grade dysplasia; HGD, high-grade dysplasia.

  • 1

    Note: Determined by endoscopic ultrasound and CT scans.

1NESurgeryNormal epithelium54MWhite   Preoperative chemoradiation
BEMetaplasia      (5-FU + cisplatin)
EACAdenocarcinoma   Moderately differentiatedT3N1III 
2NEBiopsyNormal epithelium68MWhite   Unknown
BEMetaplasia       
EACAdenocarcinoma   Moderately differentiated   
3NESurgeryNormal epithelium52MWhite   No chemoradiation
BEMetaplasia with dysplasia   Moderately differentiatedT1N0I 
4NESurgeryNormal epithelium61MWhite   No chemoradiation
BEMetaplasia with focal HGD       
EACAdenocarcinoma   Poorly differentiatedT3N1M0III 
5NEBiopsyNormal epithelium67MAsian/Pacfic Islander   No chemoradiation
BEMetaplasia with LGD and HGD       
EACAdenocarcinoma   Well differentiated   
6NEBiopsyNormal epithelium82FWhite   Unknown
EACAdenocarcinoma   Poorly differentiated   
7NEBiopsyNormal epithelium77MWhite   No chemoradiation
BEMetaplasia with LGD and focal HGD       
EACAdenocarcinoma   Moderately differentiated   
8NEBiopsyNormal epithelium80MWhite   No chemoradiation
BEMetaplasia       
9NEBiopsyNormal epithelium72MWhite   No chemoradiation
BEMetaplasia       
EACAdenocarcinoma   Moderately differentiatedT2N0M11IV1 
10NEBiopsyNormal epithelium78MWhite   No chemoradiation
BEMetaplasia       
11NEBiopsyNormal epithelium56MWhite   No chemoradiation
BEMetaplasia       

High-density SNP-chip analysis

BE or EAC and their matched normal genomic DNA (100 ng) from cases 5, 7, 9 and 6, as well as from lung cancer cell line NCI-H2171 and its paired lymphoblastoid cell line NCI-BL2171, were subjected to whole genome amplification using a REPLI-g Midi Kit according to the manufacturer's protocol (Qiagen). Genomic alterations found in the unamplified aliquots of NCI-H2171 cells were also detected in the amplified aliquots as well (Supporting Information Figure S1). All genomic DNA samples (375 ng) were analyzed on GeneChip Human mapping 250 K microarrays (SNP-chip, Affymetrix, Santa Clara, CA), as described previously.12, 14 Hybridization, washing and signal detection were performed on a GeneChip Fluidics Station 400 and a GeneChip scanner 3,000, according to the manufacturer's protocols (Affymetrix). Microarray data were analyzed for the determination of both total and allele-specific copy numbers (AsCNs) using the CNAG program, as previously described,12, 14 with minor modifications. All SNPs within a given inferred LOH region were formally analyzed as “heterozygous” SNPs (see Ref.14 for mathematical details). For clustering of samples according to copy-number change and CNN-LOH status, GNAGraph software was employed. Size, position and location of genes were identified using the UCSC Genome Browser (http://genome.ucsc.edu/).

Determination of SNP sequences in cases of CNN-LOH

To validate CNN-LOH, 3 independent SNP sequences (rs2296820, rs668026 and rs2890896) at chromosome 9p were queried in case 11. The genomic region of each SNP site was amplified by genomic PCR using specific primers, and PCR products were purified and sequenced. Primer sequences were as follows: 5′-AAA TGA CCG CAC CTC TGA AG-3′ and 5′-GAG AGC GGC AAA CCA TTA GA-3′ for rs2296820, 5′-TTT GCT AGT CTC ACC ACT TGC-3′ and 5′-CCT TGC ACA TTA TAA ACT CTC GAT-3′ for rs668026 and 5′-GGA AGG GTA GGC TTC CTG AT-3′ and 5′-TCT GTG TCT TTG GTT CTT TTT CA-3′ for rs2890896.

Quantitative genomic and mRNA real-time PCR

Gene dosages of chromosome 9p22.3 in case 11; the ERBB2 gene in cases 5 and 9; the CDKN2A gene in cases 8 and 11; and the basonuclin 2 (BNC2) gene in 11, as well as mRNA expression levels of ERBB2 in case 9 and BNC2 in case 11, were determined by quantitative real-time PCR (iCycler, Bio-Rad, Hercules, CA) using Sybr Green. To determine relative gene dosages and mRNA levels, the chromosome 2p21 region and β-actin were measured as controls, respectively. The delta threshold cycle value (ΔCt) was calculated from the given Ct value by the formula ΔCt = (Ct sample − Ct control). Fold change was calculated as 2−ΔCt. PCR was performed using the following primer pairs: 5′-CCC TCA AAA AGT GGA GAC GA-3′ and 5′-ATT CTT GGG GCA CCT CTC TT-3′ for 9p22.3, 5′-AGT ACC TGG GTC TGG ACG TG-3′ and 5′- CTG GGA ACT CAA GCA GGA AG-3′ for ERBB2 genomic DNA, 5′-GTG CCA AAG TGC TCC TGA AGC TG-3′ and 5′-AGC AAA TCT GTT TGG AGG TCTG-3 for the CDKN2A gene, 5′-GGG GAT TCT TCT CGA TGA CA-3′ and 5′-ACT CTC AGG GTC CCC TTG TT-3′ for BNC2 genomic DNA, 5′-GGC AAT CCT GGC TGC GGA TCA AGA-3′ and 5′-ATT TCT GAA CTT CTT GGC TGC C-3′ for the 2p21 region, 5′-GGG ACT ATG TCC GAG GAT AC-3′ and 5′-AGG GTG ATG ATT TCC TCT TC-3′ for BNC2 mRNA, 5′-GTT TGA GTC CAT GCC CAA TC-3′ and 5′-CCC ACG TCC GTA GAA AGG TA-3′ for ERBB2 mRNA and 5′-CCT GGC ACC CAG CAC AAT-3′ and 5′-GCC GAT CCA CAC GGA GTA CT-3′ for β-actin mRNA.

Western blot analysis and colony assays

OE33 cells were transfected with either HA-tagged BNC2 expression vector (kindly provided from Dr. Satrajit Sinha, State University of New York at Buffalo) or empty vector. Cells were harvested, lysed in 2× sample buffer (12% glycerol, 20 mM Tris-HCl, 4% SDS, 100 mM DTT, 4 mM EDTA, 0.04% Coomassie Brilliant Blue R250) and heat denatured. Samples were subjected to SDS-PAGE followed by an electrotransfer to polyvinylidene difluoride membrane. The signals were developed with either Supersignal West Pico-Chemiluminescent or -Dura Extended Duration Substrate (Pierce Biotechnology, Rockford, IL). Anti-HA antibodies were obtained from Santa Cruz Biotechnology (Santa Cruz, CA), while anti-GAPDH was obtained from Research Diagnostics (Concord, MA).

For colony formation assays, transfected cells were cultured in the presence of 250 μg/ml G418. After 2 weeks, cells were stained with Crystal Violet Dye (0.1% dissolved in 50% methanol). To quantify the number of surviving colonies, cells were dissolved in 1.0% SDS and absorbance was measured at 600 nm.

Results

Summary of SNP-chip analysis of esophageal adenocarcinoma samples

SNP chip analysis was performed on genomic DNAs from 11 matched sets (6 of NE, BE and EAC; 1 of NE and EAC; and 4 of NE and BE) containing in total 10 BE, 7 EAC and 11 matched NE samples. Because concentrations of genomic DNA in cases 5, 7, 9 and 6 were low, these samples were subjected to whole-genome PCR amplification prior to SNP-chip analysis. To confirm the reliability of DNA from whole genome amplification, we compared it to whole DNA. Both were subjected to SNP-chip analysis using the same cell line, NCI-H2171. The genomic changes found in the unamplified DNA from NCI-H2171 cells were also detected in the whole-genome amplified samples, suggesting that whole-genome amplification is a useful approach when DNA amounts are limited (Supporting Information Figure S1). BE samples from case 3 and from matched BE and EAC samples in cases 1, 4 and 2 revealed no detectable genomic abnormalities (data not shown). In contrast, the remaining specimens (6 BEs and 4 EACs) harbored one or more genomic abnormalities (Table II and data not shown). Thus, 6 (60%) of 10 BE and 6 (57%) of 7 EAC samples contained genomic alterations, including copy-number changes and CNN-LOH. CNN-LOH is a genomic abnormality that normally cannot be detected by CGH analysis; these regions usually also contain a mutation in a key gene.

Table II. Comparison of Genomic Abnormalities Between Matched Barrett's and Adenocarcinoma Cells
CaseLocationProximalDistalSize (Mb)Status in each tissueGene(s) in the region
BEEAC
  1. Physical localization and size (Mb) are obtained from UCSC Genome Browser. If less than 5 genes are in these regions, all gene names are displayed. Note that amplification of Xp22.13–p22.12 (3.5 Mb) were found in both BE and EAC from case 9.–Abbreviations: Del, deletion; Dup, duplication (gain of copy number); Amp, amplification; CNN-LOH, copy number neutral loss of heterozygosity; –, not analyzed.

59p24.3–p13.230,91037,046,81637CNN-LOHNormal>10 genes including CDKN2A
 17q12–q21.233,823,47937,542,8903.7NormalAmp>10 genes including ERBB2, CSF3, RARA and TOP2A
76q22.33–q23.2130,053,359133,756,1203.7NormalAmp>10 genes including CTGF
 6q23.2–q23.3134,406,908-136,520,8871.1NormalAmp7 genes including MYB
 9p21.321,455,47822,597,8942.1DelNormalIFNE1, MTAP, CDKN2A, CDKN2B and DMRTA1
98p21.319,243,96820,116,0500.9NormalAmpSH2D4A, ChGn, LPL, SLC18A1 and ATP6V1B2
 8p1230,973,93433,512,5922.5NormalAmpPURG, WRN, NRG1, FUT10 and RBM13
 16p13.1311,095,28411,869,3590.8NormalAmp9 genes including SOCS1
 16p13.1212,744,24813,938,6671.2NormalAmpERCC4
 17q12–q21.234,176,03637,279,7443.1NormalAmp>10 genes including ERBB2, CSF3, RARA and TOP2A
 17q21.32–q2243,910,74448,020,8204.1NormalAmp>10 genes
 17q23.2–q23.353,943,07158,922,7525.0NormalAmp>10 genes
 Xp22.13–p22.1217,538,13421,071,1743.5AmpAmp>10 genes
67p11.254,497,22455,151,9500.7AmpSEC61G and EGFR
 11p13–p1233,540,41636,817,8063.3Amp>10 genes including LMO2
111p21.3–p11.297,590,482120,863,83323.3Dup>10 genes
 3p14.260,325,74960,953,2510.6DelFHIT
 5p15.33–p1181,94946,419,09246.3Del>10 genes
 7p21.3–q11.228,467,65668,111,88459.6Del>10 genes including EGFR
 8p23.3–p12180,56834,774,31434.6Dup>10 genes
 8q13.2–q24.369,209,296146,263,53877.1Del>10 genes including MYC
 9p24.3–p21.130,91031,806,03631.8CNN-LOH>10 genes including JAK2
 9p22.3–p22.215,773,26616,931,7191.2DelBNC2
 9p21.321,823,65921,988,7330.2DelMTAP and CDKN2A
 10q25.2–q25.3113,622,691115,912,4822.3Del>10 genes
 11p15.3–p15.110,802,69317,496,1386.7Del>10 genes
 11p14.226,845,10233,281,9526.4Del>10 genes including WT1
 11p1237,649,25639,949,1392.3DelNo known genes
 14q31.2–q32.3382,684,067106,356,48223.7Dup>10 genes including BCL11Band AKT1
 16q23.177,234,87177,398,5670.2DelWWOX
 17p13.1–p11.210,538,73619,971,5529.4Dup>10 genes
 17q12-q21.231,501,49937,358,1815.9Del>10 genes including ERBB2, CSF3, RARA and TOP2A
 17q24.2–q24.363,302,86565,729,7102.4Del>10 genes
 17q24.3–q25.165,729,71069,223,3513.5Del7 genes
 18q11.217,918,38720,052,9602.1Del>10 genes including GATA6
 18q11.2–q12.120,196,46124,997,4384.8Dup9 genes
 19q12–q13.1132,977,35939,248,4646.3Del>10 genes including CEBPA and CEBPG
 20q13.1347,968,65848,404,2690.4DelZNF313, SNAI1, Kua-UEV and CEBPB
85q15–q21.196,914,45398,821,2311.9DelRGMB and CHD1
 7p21.3–p21.110,332,01913,408,6333.1DelNDUFA4, PHF14, SCIN and ARL4A
 7p15.321,468,52023,844,2052.4Del>10 genes
 7p15.1–p14.327,732,39132,309,1604.6Del>10 genes
 7q14.138,575,53239,469,2310.9DelVPS41, POU6F2 and RALA
 7p12.151,653,69752,398,4070.7DelNo known gene
 7q21.1179,524,52681,077,7691.6DelCD36, SEMA3C and HGF
 9p24.3182,1291,212,8041.0DelDOCK8, ANKRD15, DMRT1, DMRT3 and DMRT2
 9p24.2–p24.12,240,3707,362,8585.1Del>10 genes including JAK2
 9p2310,537,88611,701,4591.2DelNo known gene
 9p21.320,989,64622,673,7921.7Del>10 genes including CDKN2A
 9p21.3–p21.224,989,70225,954,8521.0DelTUSC1
 9p21.2–p21.125,954,85231,836,5275.9Del6 genes
 9q21.33–q22.285,272,21089,347,3064.1Del>10 genes
 9p22.289,347,30690,834,7521.5DelGADD45G, DIRAS2 and SYK
 9q22.2–q22.3190,834,75292,063,7051.2DelAUH, NFIL3, ROR2, SPTLC1 and IARS
 9q22.33–q31.197,923,215102,419,8964.5Del>10 genes
 9q31.1–q31.2103,707,323107,191,7453.5Del>10 genes
 9q33.3123,643,047125,825,5982.2Del>10 genes
 21q11.2–q21.114,892,58716,494,1111.6DelNRIP1 and USP25
 21q21.117,139,47022,209,6345.1DelCXADR, BTG3, CHODL, PRSS7 and NCAM2
101p32.3–p21.254,223,221100,428,18846.2Del>10 genes
 3p14.260,589,48661,002,1160.4DelFHIT
 4p16.3–p16.23,510,5714,842,0561.3Del6 genes
 7p22.3–q22.1141,322102,675,814102.5Dup>10 genes including EGFR and HGF
 7q22.1–q36.3102,675,814158,377,13455.7CNN-LOH>10 genes including MET and BRAF
 8p23.3–p12180,56830,728,60330.5Dup>10 genes
 9p24.3–p13.330,91035,208,69335.2Del>10 genes including JAK2, BNC2 and CDKN2A
 9p13.236,443,17638,589,2972.1Del>10 genes
 10p15.3–p15.1148,9464,785,2684.6Del10 genes
 12q21.173,327,35773,956,8160.6AmpKCNC2
 12q21.1–q24.3373,979,022132,377,15158.4Del>10 genes
 13q (whole)   Dup>10 genes
 17p13.3–p11.218,90122,029,23722.0Del>10 genes
 17q25.376,868,17378,598,0591.7CNN-LOH>10 genes
 18p11.32210,0711,738,7221.5CNN-LOH8 genes
 18p11.32–q11.22,013,69320,887,31118.9Dup>10 genes
 18q11.2–q12.323,055,29939,594,16116.5Del>10 genes
 18q12.3–q21.140,929,68147,747,0056.8Dup>10 genes including SMAD2, SMAD7 and SMAD4
 18q21.2–q21.3151,216,34052,531,2521.3DelTCF4, WDR7 and WDR7
 18q21.31–q21.3252,633,15855,879,0133.2Dup>10 genes
 18q21.32–q22.156,357,31660,406,6734Dup>10 genes including BCL2
 18q22.1–q22.261,129,61865,324,3314.2DelCDH7, CDH19, TXNDC10 and DOK6
 18q22.2–q2365,568,55673,578,0128Dup>10 genes
 18q2373,598,71174,272,6710.7DelNo known gene
 18q2374,163,51576,082,4671.9CNN-LOH8 genes
 20p12.1–p11.111,928,99626,236,53514.3Del>10 genes including FOXA2
 21q11.2–q22.1114,986,47432,776,07717.8Del>10 genes
 21q22.12–q22.335,141,43646,885,63911.7Del>10 genes

Chromosomal abnormalities in matched BE and EAC specimens

Since BE is believed to constitute an early step in EAC development, we compared the genomic status of BE and EAC with that of NE from the same individuals (Fig. 1). Three cases (1, 4 and 2) contained no genomic changes in either their BE or their EAC samples. BE tissue from case 9 contained 1 amplification on the X chromosome (Xp22.13-p22.12, 3.5 Mb). Interestingly, EAC tissue from this patient exhibited 7 additional amplified regions, including 8p21.3 (0.9 Mb), 8p12 (2.5 Mb), 16p13.13 (0.8 Mb), 16p13.12 (1.2 Mb), 17q12–q21.2 (3.1 Mb), 17q21.32–q22 (4.1 Mb) and 17q23.2–q23.3 (5.0 Mb), as well as Xp22.13–p22.12 (3.5 Mb). The EAC sample from case 5 contained an amplification of 17q12–q21.2 (3.7 Mb), whereas the EAC from case 7 had 2 amplifications at chromosome 6q [6q22.33–q23.2 (3.7 Mb) and 6q23.2–q23.3 (1.1 Mb)]. These amplifications were not detected in matching BE samples. Unexpectedly, CNN-LOH of 9p-terminal, p13.2 (37.0 Mb), occurred in BE tissue from case 5, as did deletion of 9p21.3 (2.1 Mb) in BE tissue from case 7; however, these aberrations were not detected in their matched EAC tissues, possibly because contamination with normal cells masked this abnormality in the frank tumors.

Figure 1.

Summary of genomic abnormalities in esophageal samples. Genomic DNAs from 6 matched sets (NE, BE and EAC), 1 matched NE-EAC pair and 4 matched NE-BE pairs (in total, 10 BE and 7 EAC samples) were subjected to SNP-chip analysis; genomic abnormalities are summarized by color: pink (copy-number-neutral LOH [CNN-LOH]); green (hemizygous deletion); blue (homozygous deletion); red (duplication/amplification). Six (60%) of 10 BE (cases 5, 7, 9, 11, 8 and 10) and 4 (57%) of 7 EAC (cases 5, 7, 9 and 6) samples possessed genomic alterations.

Other CNN-LOH and copy-number changes

All other abnormalities, including CNN-LOH and copy-number changes, are displayed in Table II. Based on SNP-chip analyses, 3 cases exhibited CNN-LOH. BE samples from cases 5 and 11 had 9p CNN-LOH, while case 10 contained 4 regions of CNN-LOH, including 7q22.1–q36.3 (55.7 Mb), 17q25.3 (1.7 Mb), 18p11.32 (1.5 Mb) and 18q23 (1.9 Mb) (Fig. 1 and Table II). Deletions and duplications were found only in BE samples. Deleted regions involving 7p, 9p, 9q and 18q were found frequently, and duplicated regions involving 18q were often observed (Fig. 1 and Table II). Of interest, 1 BE sample (case 10) manifested amplification at 12q.21.1 (0.6 Mb); this region contains only 1 known gene, the Shaw-related voltage-gated potassium channel gene known as KCNC2. Furthermore, EAC samples exhibited amplifications at 6q, 8p, 11p, 16p and 17q, but no duplications or CNN-LOH (Fig. 1 and Table II).

Shared abnormalities found in BE and/or EAC samples

Several chromosomal abnormalities were shared by 2 or more different patients. As shown in Table III, several chromosomal abnormalities were present in BE and EAC samples, suggesting that these loci harbored aberrant genes altered early during BE-EAC evolution. Shared abnormalities included 2 BE cases (5 and 11) with 9p CNN-LOH; 2 BE cases (10 and 11) with duplication of 8p23.3–p12; 2 BE cases (8 and 10) with deletion of 21q11.2–q21.1. Two EAC samples (5 and 9) exhibited amplification of 17q12–q21.2; this was the only region with gains in more than 1EAC sample. Furthermore, the CDKN2A gene was deleted in 4 BE (8, 11, 7 and 10) and the FHIT gene was deleted in 2 BE (10 and 11) samples.

Table III. Shared Abnormalities Found in Either Be and/OR EAC Samples
ChromosomeSizeStatusCase
  • Six regions were identified as shared abnormalities including CNN-LOH, duplication, deletion and amplification.–Abbreviations: CNN-LOH, copy-number neutral loss of heterozygosity; BE, Barrett's esophagus; EAC, esophageal adenocarcinoma.

  • 1

    Homozygous deletion.

9p24.3–p21.131.7 MbCNN-LOH5 BE
   11 BE
8p23.3–p1230.5 MbDuplication10 BE
   11 BE
21q11.2–q21.11.6 MbDeletion8 BE
   10 BE
17q12–q21.23.1 MbAmplification5 EAC
   9 EAC
CDKN2A geneDeletion8 BE1
   11 BE1
   7 BE
   10 BE
FHIT geneDeletion10 BE
   11 BE

Validation of SNP-chip data using nucleotide sequencing quantitative real-time PCR and expression analysis

We validated CNN-LOH detected by SNP-chip using several different techniques including determining SNP sequences, as well as gene-dosage, in the CNN-LOH region. It was presumed that when LOH occurred, SNP sequences in this region should exhibit homozygosity, whereas those of matched normal samples should be heterozygous. Therefore, we examined 3 independent SNP sequences on chromosome 9p in the CNN-LOH region in case 11 (NE and BE; Fig. 2a). In the BE sample, all 3 SNP sites (rs2296820, rs668026 and rs2890896) clearly showed only a single signal (Fig. 2b). In contrast, these 3 SNP sites in the matched NE sample showed heterozygosity. These results strongly suggested that LOH occurred in this region.

Figure 2.

Representative SNP-chip analysis in Barrett's esophagus samples (a) SNP-chip data for chromosome 9 in BE samples (cases 8 and 11). Red dots are SNP sites as probes and indicate total copy number (CN). Blue line is an average of copy number and shows gene dosage. Green bars are heterozygous (hetero) SNP calls. Red and green lines show allele-specific copy number (AsCN). Both cases exhibited deletions of CDKN2A (boxed regions). Case 11 had CNN-LOH at 9p and a deletion of the basonuclin 2 gene (circled). (b) SNP sequences in the 9p CNN-LOH region for validation of CNN-LOH in case 11. Three independent SNP sites (rs2296820, rs668026 and rs2890896) were sequenced. All 3 SNP sites showed heterozygosity in matched NE samples; however, their corresponding BE samples showed homozygosity, consistent with loss of heterozygosity. (c) Determination of gene dosage in the 9p region. Gene dosage of 9p22.3 (the CNN-LOH region) in case 11 was compared between matched NE and BE genomic DNA, using quantitative genomic real-time PCR. Gene dosage levels were calculated as the ratio between 9p22.3 and the reference genomic DNA, 2p21. Results represent the mean of 3 experiments ± SD.

Next, we determined gene dosage in this region to exclude the possibility of a hemizygous deletion. Gene dosage at 9p22.3 in the region of CNN-LOH in BE case 11 was compared with that in the matched NE using quantitative genomic real-time PCR (Q-PCR). Gene dosage levels were calculated as the ratio between 9p22.3 and the reference genomic DNA at 2p21, which displayed normal gene dosage by SNP-chip analysis. DNA levels at the 9p22.3 region in BE case 11 were almost identical to those in matched NE at this site, indicating that this region possessed normal copy number (Fig. 2c). Thus, our SNP sequence and genomic Q-PCR data validated the results of SNP-chip analysis, clearly showing CNN-LOH at the 9p region.

We also validated copy-number changes. Four BE samples (8, 11, 7 and 10) showed deletion of the CDKN2A gene by SNP-chip analysis (Fig. 2a and Table III). Gene dosage of CDKN2A was examined by genomic Q-PCR. Levels of CDKN2A in BE samples from cases 8 and 11 were approximately 10-fold lower relative to matched NE (data not shown). EAC samples from cases 5 and 9 exhibited amplification of 17q12–q21.2 by SNP-chip analysis; and this region contains the ERBB2, CSF3, RARA and TOP2A genes. Genomic Q-PCR revealed that levels of ERBB2 in EAC samples 5 and 9 were ∼5-fold and 8-fold higher than in matched NE samples, respectively (Fig. 3a). Taken together, these results validated the observations of our SNP-chip analysis.

Figure 3.

Relationship between copy-number change and gene expression (a) Amplification of the ERBB2 gene: Two EAC samples (5 and 9) had amplification of the 17q12–q21.2 region, which includes the ERBB2 gene. Gene dosage of the ERBB2 gene was compared between matched NE, BE and EAC genomic DNA using genomic Q-PCR. Gene dosage levels were calculated as the ratio between the ERBB2 gene and the reference genomic region, 2p21. Results represent the mean of 3 experiments ± SD. (b) Expression of ERBB2 mRNA in case 9 was compared between matched NE, BE and EAC samples using qRT-PCR. Expression levels were calculated as the ratio between ERBB2 and the reference gene, β-actin. Results represent the mean of 3 experiments ± SD.

Furthermore, we compared the copy-number change to gene expression levels. Expression of ERBB2 mRNA was examined with quantitative reverse transcriptase PCR (qRT-PCR) in case 9, which showed amplification of this gene by SNP-chip and genomic Q-PCR. Levels of ERBB2 mRNA in the BE and EAC from case 9 were ∼6-fold and 14-fold higher than in matched NE, respectively (Fig. 3b). Taken together, these data suggested that copy-number changes resulted in aberrant gene expression.

Deletion of basonuclin 2 in Barrett's sample

As shown in Figure 2a, the BE sample from case 11 clearly showed deletion involving the 9p22.3–p22.2 region (1.2 MB, circled); only 1 known gene, basonuclin 2 (BNC2), is located here. BNC2, a zinc-finger protein, can bind to DNA and behave as a transcription factor.25–27 Genomic Q-PCR revealed that the level of BNC2 in BE sample 11 was ∼3.5-fold lower than in matched NE (Fig. 4a). The level of BNC2 mRNA in this BE sample was also ∼3.5-fold lower than in matched NE (Fig. 4b), consistent with decreased expression of BNC2 in this BE sample due to loss of this gene.

Figure 4.

Deletion of basonuclin 2 in Barrett's esophagus. (a) Deletion of the basonuclin 2 (BNC2) gene: BE sample 11 exhibited deletion at 9p22.3–p22.2; the deleted region contained only the BNC2 gene. Gene dosage of BNC2 in the BE sample was compared to matched NE genomic DNA using genomic Q-PCR. Gene dosage levels were calculated as the ratio between the BNC2 gene and the reference genomic region, 2p21. Results represent the mean of 3 experiments ± SD. (b) Expression of BNC2 mRNA in case 11 BE was compared to its matched NE sample using qRT-PCR. Expression levels were calculated as the ratio between BNC2 and the reference gene, β-actin. Results represent the mean of 3 experiments ± SD.

These findings prompted us to examine forced expression of BNC2 in the EAC cell line, OE33. Clonogenic assays were used to assess the effect of BNC2 on the growth rate of OE33 cells, since these cells do not express endogenous BNC2 mRNA (data not shown). OE33 cells were transfected with either a BNC2 expression vector or an empty vector as a control. Each vector contained the neomycin resistance gene. Untransfected cells all died in 250 μg/ml G418 after 2 weeks. Therefore, cells were cultured in media containing 250 μg/ml G418 for 2 weeks and then stained to determine the number of surviving colonies. BNC2-transfected cells formed 60% fewer colonies than did empty vector-transfected controls (Figs. 5b and 5c). Taken together, these results suggested that BNC2 inhibits clonal proliferation of EAC cells and is a potential EAC tumor suppressor gene.

Figure 5.

Inhibition of proliferation of esophageal adenocarcinoma cells by BNC2. (a) Expression of exogenous HA-tagged BNC2. OE33 esophageal adenocarcinoma cells were transfected with either a BNC2 expression vector or an empty vector as a control. Expression of HA-tagged BNC2 was determined by Western blot analysis. GAPDH was used as an internal control. (b) Colony formation assay. Transfected cells were cultured in the presence of 250 μg/ml G418 for 2 weeks and then stained with 0.1% crystal violet to determine the number of surviving colonies. (c) Quantification of the number of surviving colonies. The colonies were dissolved in 1.0% SDS and absorbance was measured at 600 nm. Results were significantly different between control and BNC2-transfected cells (p < 0.001).

Discussion

In our study, we found that 6 (60%) of 10 BE and 4 (57%) of 7 EAC samples contained genomic alterations, including both copy-number changes and CNN-LOH. Analysis of matched NE, BE and EAC sample sets revealed several chromosomal regions potentially involved in the progression from BE to EAC. Furthermore, we identified a zinc-finger protein, BNC2, as a potential EAC tumor suppressor gene.

Our analyses identified deletions of 9p (4 cases) and 18q (2 cases) in BE samples and amplification of 17q (2 cases) in EAC samples. These findings are consistent with previous EAC results obtained by CGH analysis.4–11 We also found that the 17q12–q21.2 region was amplified in 2 EAC samples; this region includes the ERBB2, CSF3, RARA and TOP2A genes. ERBB2 is a transmembrane glycoprotein with tyrosine kinase activity, for which 10–70% of EAC samples show amplification.28–34 An antagonist of ERBB2, Trastuzumab/Herceptin, inhibits growth of the OE19 EAC cell line, which exhibits high expression of ERBB2.35 DNA topoisomerase II alpha (TOP2A) is associated with active cell proliferation, and its overexpression has been reported in a number of tumors, including esophageal squamous cell carcinoma.36, 37 An inhibitor of topoisomerase II, etoposide, is used for chemotherapy of solid tumors including small-cell lung cancer. Data suggest that breast cancers with amplification of ERBB2 and TOP2A have a better response when they receive the combination of both Trastuzumab/Herceptin and a TOP2A inhibitor.38 In reference to our observed amplification of RARA, it is notable that retinoic acid, including all-trans retinoic acid (ATRA) and 9-cis retinoic acid (9-cis RA), induces the differentiation of acute promyelocytic leukemia cells and neuroblastoma cells and is used in the treatment of these and other cancers.

Our SNP-chip analysis showed that the EAC from case 7 exhibited amplification of 6q22.33–q23.2 including the CTGF gene. This gene is involved in cell adhesion, migration, proliferation and angiogenesis.39 A recent study by our group demonstrated that 75% of esophageal squamous cell carcinomas overexpressed CTGF with accumulation of β-catenin in the nucleus,40 supporting the importance of the β-catenin signaling pathway in esophageal tumorigenesis. In case 9, the BE sample contained only 1 chromosomal amplification (Xp22.13–p22.12), whereas the EAC sample from the same patient had numerous abnormalities including X-chromosomal alteration. This finding suggests that X-chromosomal alteration was acquired as an early event; this alteration may enhance the evolution from BE to EAC.

We also identified several CNN-LOH regions that included the entire 9p arm, 7q22.1–q36.3 (55.7 Mb), 17q25.3 (1.7 Mb), 18p11.32 (1.5 Mb) and 18q23 (1.9 Mb). CNN-LOH is a genomic abnormality that cannot normally be detected by karyotypic analysis; CNN-LOH regions often contain mutated genes. For example, the constitutively active forms of either JAK2 V617F mutant, FLT3-ITD, or an AML1/RUNX1 frameshift were found in a CNN-LOH region in AML cells from our SNP-chip analysis.24 The 7q22.1–q36.3 CNN-LOH region in BE contains the proto-oncogenes MET and BRAF. An activating mutation of MET (Y1253D) was detected in 15 (11%) of 138 patients with oropharyngeal squamous cell carcinoma.41 An activating mutation of BRAF (V600E) was found in 66% of malignant melanomas, as well as at lower frequencies in a wide range of human cancers42; in particular, 11% of Barrett's EACs have been reported to possess BRAF mutations.43 These findings prompted us to determine the exon sequences at mutational hotspots (MET Y1253 and BRAF V600) in sample 10; however, these genes did not contain detectable mutations (data not shown). Thus, further studies are indicated to identify key dysregulated or abnormal gene(s) in these regions.

In cancer cells, tumor suppressor genes are often inactivated by deletion, mutation and/or hypermethylation of their promoter regions. The cyclin-dependent kinase inhibitor gene, CDKN2A, is homozygously or hemizygously deleted, mutationally inactivated or hypermethylated in ∼50, 5 and 60% of EACs, respectively, and hypermethylated or hemizygously deleted in up to 40% of BEs.44–49 These findings are consistent with our identification of CDKN2A deletions in 4 BE samples (40%) and emphasize that CDKN2A inactivation represents an early event in the BE-EAC carcinogenic cascade. Similarly, the FHIT gene was deleted in 2 BE samples (20%). An earlier report observed that 86% of BE and 93% of EAC samples exhibited altered FHIT expression.50 Although aberrant FHIT expression has also been detected in normal tissues,51FHIT-deficient mice developed tumors in several tissues, suggesting that this gene functions as a tumor suppressor.52, 53

The frequency of chromosomal changes in our study differed slightly from previously published CGH results; this discrepancy may have been due to high normal cell contamination of several of our samples or if several were triploidy and tetraploidy. This can be difficult to identify by SNP-chip analysis. Surprisingly, chromosomal alterations sometimes differed between BE and EAC samples. BE samples from cases 5 and 7 contained chromosomal changes in the 9p region, but these alterations disappeared in the corresponding EAC samples. One possible explanation for this conundrum is that the 9p alteration-bearing population was dominant in the BE sample; expansion of this clone altered the cellular microenvironment, thus enhancing the growth of a completely different (i.e., 9p-euploid) cellular population, which then became dominant in the EAC sample. Such clonal evolution has been previously reported in Barrett's neoplastic evolution.54–58

Interestingly and to our knowledge for the first time, homozygous deletion of the BNC2 gene occurred in 1 BE sample. BNC2 is a zinc-finger protein that is highly expressed in normal keratinocytes, ovary, testis, kidney and lung.25–27 BNC2 can bind to a sequence in the promoter region of the rRNA gene, and its protein is localized to the nucleus, suggesting that it functions as a transcriptional regulator.25–27 Also interestingly, the esophageal adenocarcinoma cell lines OE33 and OE19 did not express this gene (data not shown), and overexpression of BNC2 inhibited clonal proliferation of OE33 cells, suggesting that this gene has potential tumor-suppressive function. Inactivation by methylation and/or deletion of several tumor suppressor genes, including CDKN2A and TP53, are known to be involved in esophageal tumorigenesis. Expression of BNC2 was low in a BE sample and the EAC cell lines. This inactivation could contribute to transformation to esophageal cancer cells. Recently, Nancarrow et al.59 analyzed 23 EAC samples by SNP-chip. These investigators found several chromosomal lesions that were similar to our data, including homozygous deletions of CDKN2A and FHIT, loss of copy number at 17p, and CNN-LOH at 17q25.3, further emphasizing the importance of these regions or genes in the development of EAC.

In summary, we have demonstrated that BE frequently exhibits certain genomic and chromosomal alterations consistent with its precancerous state, often with early timing of these events. Furthermore, we have identified several novel genomic abnormalities in BE or EAC, notably CNN-LOH and inactivation of the zinc-finger gene, BNC2.

Acknowledgements

We thank the members of laboratories for helpful discussions. We are grateful to Dr. Satrajit Sinha, State University of New York at Buffalo, for his gift of the BNC2 expression vector. S.J.M. holds the Harry and Betty Myerberg/Thomas R. Hendrix Endowed Professorship in Gastroenterology and is a member of the Sidney Kimmel Comprehensive Cancer Center and the Cellular and Molecular Medicine Graduate Program at JHU. H.P.K. is the holder of the Mark Goodson endowed Chair in Oncology Research and is a member of the Jonsson Cancer Center and the Molecular Biology Institute, UCLA.

Ancillary