Monochromosome Transfer and Microarray Analysis Identify a Critical Tumor-Suppressive Region Mapping to Chromosome 13Q14 and THSD1 in Esophageal Carcinoma

Monochromosome Transfer and Microarray Analysis Identify a Critical Tumor-Suppressive Region Mapping to Chromosome 13q14 and THSD1 in Esophageal Carcinoma

Josephine M.Y. Ko,1 Pui Ling Chan,1 Wing Lung Yau,1 Ho Kin Chan,1 King Chi Chan,1 Zhuo You Yu,1 Fung Mei Kwong,1 Lance D. Miller,6 Edison T. Liu,6 Li Chun Yang,1 Paulisally H.Y. Lo,1 Eric J. Stanbridge,7 Johnny C.O. Tang,2,3 Gopesh Srivastava,3 Sai Wah Tsao,4 Simon Law,5 and Maria L. Lung1

1Department of Biology and Center for Cancer Research, Hong Kong University of Science and Technology; 2Department of Applied Biology and Chemical Technology, Hong Kong Polytechnic University; Departments of 3Pathology, 4Anatomy, and 5Surgery, University of Hong Kong, Hong Kong (Special Administrative Region), People’s Republic of China; 6Cancer Biology and Pharmacology, Genome Institute of Singapore, Biomedical Sciences Institutes, Singapore; and 7Department of Microbiology and Molecular Genetics, University of California, Irvine, California

Abstract 100% of ESCC and other cancer cell lines. Mechanisms Loss of chromosome 13q regions in esophageal for THSD1 silencing in ESCC involved loss of squamous cell carcinoma (ESCC) is a frequent event. heterozygosity and promoter hypermethylation, as Monochromosome transfer approaches provide analyzed by methylation-specific PCR and clonal direct functional evidence for tumor suppression by bisulfite sequencing. Transfection of wild-type THSD1 chromosome 13 in SLMT-1, an ESCC cell line, and into SLMT-1 resulted in significant reduction of identify critical regions at 13q12.3, 13q14.11, and colony-forming ability, hence providing functional 13q14.3. Differential gene expression profiles of three evidence for its growth-suppressive activity. These tumor-suppressing microcell hybrids (MCH) and findings suggest that THSD1 is a good candidate TSG. their tumorigenic parental SLMT-1 cell line were (Mol Cancer Res 2008;6(4):592–603) revealed by competitive hybridization using 19k cDNA oligonucleotide microarrays. Nine candidate 13q14 tumor-suppressor genes (TSG), including RB1, showed Introduction down-regulation in SLMT-1, compared with NE1, an Esophageal cancer is geographically diverse, with only a immortalized normal esophageal epithelial cell line; 10.7% 5-year survival rate (1). Esophageal squamous cell carci- their average gene expression was restored in MCHs noma (ESCC) is the major histologic form. ESCC molecular compared with SLMT-1. Reverse transcription-PCR pathogenesis still remains poorly understood. validated gene expression levels in MCHs and a Chromosome 13q deletions are frequent events in several panel of ESCC cell lines. Results suggest that the human cancers, including ESCC (2, 3), nasopharyngeal (4), and tumor-suppressing effect is not attributed to RB1, lung (5) cancers. This current study is the first functional study but instead likely involves thrombospondin type I of the tumor-suppressive role of chromosome 13 in ESCC and domain-containing 1 (THSD1), a novel candidate TSG is initiated by the high-frequency ESCC chromosome 13q mapping to 13q14. Quantitative reverse transcription- losses detected by comparative genomic hybridization (3) and PCR detected down-regulation of THSD1 expression in loss of heterozygosity (LOH) studies (2, 6-9). Comparative genomic hybridization analysis showed 100% losses on chromosome 13q in 17 ESCC cases (3). The extremely high 13q loss was independently verified in genome-wide LOH Received 4/4/07; revised 11/4/07; accepted 12/20/07. studies (2), with 95% of 77 13q markers showing LOH. Thus, Grant support: Research Grants Council of the Hong Kong Special Administration Region, China (M.L. Lung), for The Hong Kong University of functional inactivation of TSGs on chromosome 13q is likely Science and Technology 6415/06M grant. key to ESCC development. The costs of publication of this article were defrayed in part by the payment of Monochromosome transfer into tumorigenic cell lines allows page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. functional complementation of existing defects and study of Note: Supplementary data for this article are available at Molecular Cancer tumor-suppressive effects driven by native endogenous regula- Research Online (http://mcr.aacrjournals.org/). tory environments, with control of single gene copy number Requests for reprints: Maria Li Lung, Department of Biology and Center for Cancer Research, Hong Kong University of Science and Technology, Clear Water gains more closely mimicking normal physiologic levels. Our Bay, Kowloon, Hong Kong (Special Administrative Region), People’s Republic previous studies identified tumor-suppressive regions localized of China. Phone: 852-2358-7307; Fax: 852-2358-1559. E-mail: [email protected] Copyright D 2008 American Association for Cancer Research. to 3p14 (10), 9q33-34 (11), and 14q32 (12) in the ESCC cell doi:10.1158/1541-7786.MCR-07-0154 line SLMT-1. This present study examines the tumor-suppressive

592 Mol Cancer Res 2008;6(4). April 2008 Downloaded from mcr.aacrjournals.org on September 26, 2021. © 2008 American Association for Cancer Research. 13q14 Critical Region and THSD1 in Esophageal Cancer 593

FIGURE1. Chromosome 13 allelotyping of SLMT-1 with 50 markers revealed two regions of contiguous homozygosity in SLMT-1, one with 10 markers at 13q12.11-13q12.3 and the other with 27 markers from the 13q13.3 region to the telomeric end of chromosome 13q. Microsatellite typing – deletion mapping analysis of chromosome 13 MCHs and TSs delineated four CRs within regions 13q12.3 (CR1 at D13S1299-D13S1229 and CR2 at D13S1226), 13q14.11 (CR3 at D13S263), and 13q14.3 (CR4 at D13S133, which is 0.292 Mb from THSD1) due to nonrandom loss at specific markers in the TSs. The presence (o), endogenous loss ( ), and exogenous loss (.) of markers are depicted. 1, homozygous allelic pattern; 2, heterozygous alleles are present in NE1, EC18, KYSE180, and SLMT-1;*, genes mapping to this region.

role of chromosome 13 in SLMT-1. Comparative differential 13 transfer into MCH13-111, MCH13-113, and MCH13-117 gene expression observed after competitive hybridization in (Supplementary Fig. S1). cDNA oligonucleotide microarrays of the tumorigenic parental PCR microsatellite typing confirmed the presence of donor cell line and three tumor-suppressing chromosome 13 micro- and recipient alleles, validating the successful transfer of cell hybrids (MCH) identified novel candidate TSGs. chromosome 13 into SLMT-1 in three MCH13 cell lines. Of Thrombospondin type I domain-containing 1 (THSD1), at 50 microsatellite markers, 37 were informative. An ideogram 13q14.3, showed 100% down-regulation in a panel of 18 ESCC summarizes the genotyping results (Fig. 1). and other cancer cell lines. Transfection of wild-type THSD1 into SLMT-1 significantly reduced colony formation, provid- Tumorigenicity Assay of MCH Cell Lines ing functional evidence for growth-suppressive activity. Statistically significant delays in tumor growth kinetics and The current data indicate that the mechanism for THSD1 reduced tumor sizes were observed with all three chromosome down-regulation in ESCC involved both LOH and epigenetic 13 hybrids compared with the parental cell line (Fig. 2A). The silencing. Demethylation treatment restored THSD1 expression tumorigenic potentials of MCH13-111 and MCH13-113 in THSD1 down-regulated ESCC cell lines; results from methylation-specific PCR (MSP) analysis of ESCC cell lines and bisulfite sequencing of the promoter region of THSD1 in both cell lines and primary tissues showed that loss of THSD1 expression could be partially attributed to hypermethylation in ESCC.

Results Chromosome 13 Allelotyping of Cell Lines Fifty markers were used for allelotyping ESCC SLMT-1 and human chromosome 13 donor cell line, MCH204.3 (Fig. 1). SLMT-1 contained two regions of contiguous homozygosity, one with 10 markers at 13q12.11-13q12.3 and another with 27 markers at the 13q13.3 region to the telomeric end of chromosome 13q. The two consecutive homozygous regions only contain a single allele in each of the 37 loci studied. The random chance for this occurring in a diploid genome is small. These results suggest nearly a complete deletion of a single copy of chromosome 13 in SLMT-1 cells. The loss of 13q was not only observed molecularly by allelotyping but also by comparative genomic hybridization analysis (data not shown). Loss of one copy of nearly the entire chromosome 13 in SLMT-1 strongly suggested the presence of TSGs, based on the assumption that the relevant genes in the remaining allele would also be inactivated.

Transfer of Human Chromosome 13 into an ESCC Cell Line Fluorescence in situ hybridization (FISH) of SLMT-1 using human chromosome 13 WCP probe showed one intact signal of chromosome 13 and six signals representing translocations to other chromosomes (Supplementary Fig. S1). None of the morphologies of chromosome 13 detected in the metaphase spreads appeared normal. Using microcell-mediated chromosome transfer, eight chromosome 13 MCH cell lines were obtained. Screening by DNA slot blot hybridization confirmed that five MCHs were mouse DNA-free (data not shown). FIGURE2. A . Tumor growth kinetics of the tumorigenic recipient, SLMT-1, were compared with chromosome 13 MCHs: MCH13-111, FISH and Microsatellite Typing Analysis of Chromosome MCH13-113, and MCH13-117. Points, average volume of six inoculation sites. All three chromosome 13 MCH cell lines were tumor suppressive. 13 MCHs The tumor growth kinetics of the chromosome 13 MCHs were compared The distinctive acrocentric pattern of the transferred exo- with their corresponding tumor segregants, MCH13-111/TS2R and genous chromosome 13 in the metaphases of chromosome 13 MCH13-117/TS1L. B. Representative results of D13S133 microsatellite typing at CR4 for the recipient SLMT-1, donor MCH204.3, hybrid MCH13- MCHs enabled its precise identification. Whole chromosome 111, and tumor segregant MCH13-111/TS2R. w, exogenous donor allele 13 FISH analysis verified successful exogenous chromosome transfer; !, allelic loss.

Mol Cancer Res 2008;6(4). April 2008 Downloaded from mcr.aacrjournals.org on September 26, 2021. © 2008 American Association for Cancer Research. 13q14 Critical Region and THSD1 in Esophageal Cancer 595

differed significantly from the SLMT-1 recipient cell line KIAA0853, ESD, RB1, CHC1L, PHF11, RFP2, FLJ11712, (P < 0.001). For MCH13-117, only three of six sites formed THSD1, and C13orf9, mapped to chromosome 13q14. Their tumors with significantly longer latency periods of 2 to 8 weeks average expression values, detailed physical locations, descrip- (P = 0.001). MCH13-111 formed tumors in four of six sites tions, and gene ontologies are shown in Table 1A. with latency periods of 3 to 5 weeks. MCH13-113, with dele- A list of the remaining 40 genes on other chromosomes, tion of 13q22.1-qter, showed uniformly smaller tumors and which may be involved in the downstream cascades of tumor latency periods of 7 to 18 weeks. This clone showed that suppression in the nude mouse model, is arranged according transfer of chromosome 13q, missing the 13q22.1 region to their gene ontology terms (Table 1B). The categories of around D13S156, 13q31.1 region around D13S170, and genes involved include cell proliferation (CDKN1A, NFKBIA, 13q32.1-qter region, retains its tumor-suppressive effects. RB1, RFP2), cell death regulator (TNFAIP3), transcription factor (JUN, FOXD1, RUNX2), signal transduction (ST7, CHRNE, TORC3), extracellular matrix (TFPI2), cell adhesion Tumor Segregant Analysis (LAMC2, ITGA5, L1CAM), cytoskeleton (GAF1, ENC1, Reconstitution of mouse tumors into cell culture established PLEKHC1), immune response (DAF, IL1B, IFIT5), heat shock eight tumor segregant (TS) cell lines. Thirty-seven micro- proteins (DNAJB9, HSPA5, HSPCAL3), calcium and metal ion satellite markers spanning the whole chromosome 13 detected binding (FSTL1, CAMLG, ZSWIM6, PJA2), and others. TSdeletions in four critical regions (CR; Fig. 1). CR1 displayed 87% and 75% exogenous and endogenous allelic loss, respectively. This 2.34-Mb CR is bounded by D13S217 and Reverse Transcription-PCR Analysis of 13q14 Candidate D13S1287. The 0.373-Mb CR2 bordered by D13S893 and TSGs and Real-time Quantitative Analysis of THSD1 and D13S260 is lost in half of the TSs. CR3 is f5.99 Mb mapping PHF11 in Chromosome 13 MCHs, ESCC, and Other between D13S218 and D13S757; specific exogenous loss at Cancer Cell Lines D13S263 was detected in 50% of TS cell lines. Representative The differential gene expression of eight 13q14 candidate loss of D13S133 in CR4 in MCH13-111/TS2R is shown in TSGs in NE1, SLMT-1, and the three tumor-suppressing Fig. 2B. CR4 is f0.979 Mb between D13S1325 and D13S137; chromosome 13 MCHs was verified by reverse transcription- specific exogenous loss at D13S133 occurs in 50% of TSs. PCR (RT-PCR; Fig. 3A). An additional 15 ESCC, cervical Nude mouse assays of two TScell lines carrying different cancer (HeLa), and lung cancer (A549) cell lines show the deletions were carried out to assess their functional conse- general importance of these candidate TSGs (Fig. 3A). All quences on tumorigenicity. MCH13-111/TS2R, which exhi- eight genes were up-regulated in chromosome 13 MCHs bited the most extensive loss of the transferred chromosome 13, compared with SLMT-1 and down-regulated in SLMT-1 displayed the highest tumorigenicity (Fig. 2A). Both of the compared with NE1, as observed by the microarray analysis. tumor segregants, MCH13-111/TS2R and MCH13-117/TS1L, The most promising candidate was THSD1, which has two showed a higher tumorigenicity compared with their matched transcript variants. All 18 cell lines examined showed either MCHs (Fig. 2A), implying that multiple TSGs may reside on moderate to severe down-regulation or absence of gene chromosome 13q, which contribute to tumorigenesis. expression compared with NE1. PHF11 was another important gene detected with 100% down-regulation in 18 cell lines Gene Expression Profiling of Tumor-Suppressing MCHs with severe down-regulation or absence of gene expression in versus Tumorigenic SLMT-1 by Microarray Analysis 10 of 18 (55.6%) and slight to moderate down-regulation in 8 of Successful chromosome 13 transfer into SLMT-1 resulted in 18 (44.4%). For KIAA0853 and CHC1L, down-regulation or tumor suppression; presumably intact chromosome 13 candi- absence of gene expression was observed in 11 of 18 (61.1%) date TSG(s) are restored in those MCHs. Thus, whole genome and 9 of 18 (50.0%), respectively. A low frequency of severe differential gene expression of the three tumor-suppressing loss of expression was detected for the remaining candidates MCHs was compared with the tumorigenic SLMT-1 by gene ESD (0 of 18), RFP2 (2 of 18), FLJ11712 (0 of 18), and profiling on 19k oligonucleotide microarrays. Implicated can- C13orf9 (0 of 18). didate TSGs show down-regulated gene expression in cancer The down-regulation fold changes in THSD1 and PHF11 cells compared with normal cells. Hence, the gene expression were most significant and were further quantified by real- profile of SLMT-1 versus NE1, a reference immortalized time PCR; both showed 100% down-regulation compared esophageal epithelial cell line, was also examined by with NE1. Results were concordant with RT-PCR results competitive microarray hybridization. Genes down-regulated (Fig. 3B). in SLMT-1 compared with the immortalized normal esophageal cell line, as well as up-regulated in the three-suppressing THSD1 Transfection and Colony Formation Assay hybrids, were shortlisted as chromosome 13 candidate TSGs. The consistent down-regulation of THSD1 in all 18 cancer The 19k chip contains 248 chromosome 13 oligonucleotide cell lines and up-regulation in three tumor-suppressive hybrids probes representing f44.6% of the total possible 556 chromo- strongly suggested THSD1 as a good candidate TSG. The full- some 13 genes (National Center for Biotechnology Informa- length wild-type cDNAs of both THSD1 variants were 8 tion Map Viewer, Homo sapiens Build 36.1). Nine candidates, transfected into SLMT-1 to obtain functional evidence of their effect on growth suppression. RT-PCR showed that the THSD1 was overexpressed 1 day after transient transfection (Fig. 4A). 8 http://www.ncbi.nlm.nih.gov/ Compared with the pCR3.1 vector-alone transfectants,

TABLE1. Candidate TSGs Down-Regulated in SLMT-1 Compared with NE1and Up-Regulated in the Three Chromosome 13 MCHs Compared with SLMT-1 by Oligonucleotide Microarray Hybridization

A. Details of Candidate TSGs on Chromosome 13

Cytogenetic Gene Genbank Description/Gene Ontology Terms SLMT-1/NE1 Average MCH/SLMT-1 Average Location Accession No. Fold Difference Fold Difference

13q14.13 KIAA0853 AB020660 KIAA0853/nucleic acid binding À2.33 +1.60 13q14.2 ESD AF112219 Esterase D/formylglutathione hydrolase/hydrolase À5.09 +1.60 activity; serine esterase activity 13q14.2 RB1 NM_000321 Retinoblastoma 1/negative regulation of cell cycle; À2.70 +1.89 regulation of transcription, transcription factor activity 13q14.2 CHC1L NM_001268 Chromosome condensation 1-like/unknown À3.35 +1.65 13q14.2 PHF11 NM_016119 PHD finger protein 11/unknown À9.47 +1.72 13q14.2 RFP2 NM_005798 Ret finger protein 2/ubiquitin ligase complex; positive À4.70 +1.95 regulation of I-nB kinase/NF-nB cascade; negative regulation of cell cycle 13q14.3 FLJ11712 AK022667 Hypothetical protein FLJ11712/unknown À3.38 +2.54 13q14.3 THSD1 NM_018676 Thrombospondin, type I, domain containing 1/cell surface; À9.62 +1.62 integral to membrane 13q14.3 C13orf9 NM_016075 Chromosome 13 open reading frame 9/unknown À2.60 +1.84 B. Details of Candidate TSGs on Other Chromosomes

Gene Cytogenetic Genbank SLMT-1/NE1 Average MCH/SLMT-1 Average Location Accession No. Fold Difference Fold Difference

Cell proliferation CDKN1A 6p21.2 U03106 À7.94 +2.31 NFKBIA 14q13 NM_020529 À3.67 +2.07 Cell death regulator TNFAIP3 6q23 NM_006290 À15.47 +2.14 Transcription factor JUN 1p32-p31 NM_002228 À4.15 +1.79 FOXD1 5q12-q13 NM_004472 À6.57 +1.94 RUNX2 6p21 AL353944 À16.03 +2.55 Signal transduction ST7 8q22.2-q23.1 NM_013437 À2.56 +1.57 CHRNE 17p13-p12 Z27405 À5.14 +1.53 TORC3 15q26.1 AK025521 À4.88 +2.05 Extracellular matrix TFPI2 7q22 NM_006528 À3.63 +2.43 Cell adhesion LAMC2 1q25-q31 NM_005562 À7.10 +1.70 ITGA5 12q11-q13 NM_002205 À15.41 +1.54 L1CAM Xq28 NM_000425 À7.16 +3.11 Cytoskeleton GAF1 2p13-p12 AB020664 À7.44 +2.13 ENC1 5q12-q13.3 NM_003633 À4.05 +1.92 PLEKHC1 14q22.1 Z24725 À8.92 +1.67 Immune response DAF 1q32 M30142 À3.59 +3.62 IL1B 2q14 M15330 À19.20 +3.22 IFIT5 10q23.32 NM_012420 À5.77 +1.63 Heat shock proteins DNAJB9 7q31 NM_012328 À5.94 +1.94 HSPA5 9q33-q34.1 AF216292 À5.84 +2.20 HSPCAL3 11p14.2-p14.1 M30627 À5.69 +1.59 Calcium ion binding FSTL1 3q13.33 AK025860 À5.52 +3.31 CAMLG 5q23 NM_001745 À3.00 +1.61 Metal ion binding ZSWIM6 5q12.1 AB046797 À4.76 +1.76 PJA2 5q22.1 NM_014819 À3.20 +1.56 Miscellaneous/unknown CGI-49 1q44 NM_016002 À3.31 +1.47 SLC4A3 2q36 NM_005070 À5.67 +1.38 PHLDB2 3q13.13 AL137663 À3.27 +2.07 RAI16 8p21.3 AK025454 À2.77 +1.87 KIAA0711 8p23.3 NM_014867 À3.44 +1.70 GRINA 8q24.3 AL157442 À2.73 +1.68 TDRD7 9q22.33 AB025254 À2.77 +1.29 TRIM8 10q24.3 AF086326 À4.76 +1.96 MGC4266 12p13.33 AK021694 À5.12 +1.96 ARK5 12q24.11 NM_014840 À3.89 +1.39

(Continued on the following page)

TABLE1. Candidate TSGs Down-Regulated in SLMT-1 Compared with NE1and Up-Regulated in the Three Chromosome 13 MCHs Compared with SLMT-1 by Oligonucleotide Microarray Hybridization (Cont’d)

B. Details of Candidate TSGs on Other Chromosomes

Gene Cytogenetic Genbank SLMT-1/NE1 Average MCH/SLMT-1 Average Location Accession No. Fold Difference Fold Difference

Miscellaneous/unknown GDF15 19p13.1-13.2 NM_004864 À5.37 +6.58 DDX39 19p13.13 NM_005804 À3.34 +1.41 TMEPAI 20q13.31-q13.33 NM_020182 À9.80 +2.29 UTP14A Xq26.1 NM_006649 À3.22 +2.92 a significant reduction of colony numbers was observed in aberrant methylated CpG sites. In the ESCC 80N nontumor THSD1 transfected cells (Fig. 4B). There was a 94% and tissue, no methylated sites were detected; the ESCC tumor 80T 93% reduction of colonies with THSD1 v1 and v2, respectively tissue showed heterogeneous methylation. Heterogeneous (P < 0.001). methylation was also observed with both the 2N nontumorous tissue and the 2T tumor specimens, but the level of methylation in the tumor tissue was very extensive compared Allelotyping of ESCC Cell Lines Indicates Genomic Loss with the normal tissue. The two paired normal and ESCC around the THSD1 Region tumor specimens (80N/T and 2N/T), which showed down- Based on the LOH and comparative genomic hybridization regulation of THSD1 by RT-PCR (data not shown), were data from ESCC studies, one copy of chromosome 13 may be included for bisulfite sequencing analysis to rule out any lost in ESCC cell lines. To investigate this possibility, possibility of cell line artifacts. The MSP and bisulfite microsatellite typing of NE1, EC18, and KYSE180 with 11 sequencing data show that the THSD1 promoter is partially markers spanning 24.715 Mb from D13S757 (43.921 Mb) to and heterogeneously hypermethylated in ESCC cell lines and D13S275 (68.636 Mb) on chromosome 13 was done. NE1, primary tumor tissues. which served as the control cell line, was homozygous in 2 (D13S887 and D13S1320) of 11 markers (Fig. 1). The result THSD1 was typical for cell lines carrying two copies of chromosome Restoration of Expression in ESCC Cell Lines by ¶ 13, because the microsatellite markers are highly polymorphic. 5-Aza-2 -Deoxycytidine and Trichostatin A Treatment However, all 11 markers (f24.715 Mb) were consecutively Further evidence of promoter methylation regulation of homozygous for EC18. Except for the telomeric D13S275 THSD1 gene expression was shown by a substantial reactiva- marker, 10 of 11 markers (f18.781 Mb) were also consecu- tion of THSD1 expression in EC18 and KYSE180 with a ¶ tively homozygous for KYSE180 (Fig. 1). These results imply demethylation drug, 5-aza-2 -deoxycytidine (Fig. 4E). With the ¶ that at least a large portion of one of the two copies of treatment of both 5-aza-2 -deoxycytidine and trichostatin A, the chromosome 13, which spans the THSD1 region at 51.849 Mb, THSD1 (variant 1) gene expression was fully reactivated to was lost in EC18 and KYSE180. the level of NE1, as shown by quantitative RT-PCR (Fig. 4F). The significance of other indirect transcriptional activation and epigenetic modifications such as histone deacetylation Methylation Status of THSD1 Promoter in Immortalized in silencing of THSD1 was shown by the synergistic effect Esophageal Epithelial and ESCC Cell Lines and Tumors of restoration of THSD1 expression in the presence of both The THSD1 gene expression levels in ESCC cell lines were 5-aza-2¶-deoxycytidine and trichostatin A, an inhibitor of down-regulated up to 600-fold (Fig. 3B). In addition to LOH, histone deacetylase (Fig. 4E and F). the reduced THSD1 gene expression was hypothesized to be due to epigenetic modification of the THSD1 gene by promoter hypermethylation and/or histone deacetylation. To Discussion ascertain whether promoter hypermethylation is one of the Introduction of a single intact human chromosome 13q via mechanisms for THSD1 silencing, MSP analysis of the microcell-mediated chromosome transfer resulted in a marked THSD1 promoter region was done in six ESCC cell lines decrease in tumorigenicity of SLMT-1. Small tumors arose in versus an immortalized esophageal normal epithelial cell line, the mice only after a delayed latency period. Cell lines NE1. Only the unmethylated allele was detected in the control established from these tumor segregants, when reinjected into NE1, whereas both the methylated and unmethylated alleles mice, showed an increased tumorigenic potential (Fig. 2A). To were detected in all six ESCC cell lines (Fig. 4C). localize the tumor-suppressive activities, microsatellite deletion Additionally, clonal bisulfite sequencing of NE1, KYSE140, mapping was done to identify critical regions of nonrandom KYSE180, and two pairs of matched nontumor and ESCC losses in the TSs derived from the suppressive MCHs. CRs tumor tissues (2N/T and 80N/T) was done. The methylation were localized at 13q12 and 13q14. status of 12 CpG sites in the THSD1 promoter obtained by bisulfite sequencing is summarized in Fig. 4D. The results CRs at 13q12.3 show that NE1 does not contain methylated CpG sequences, The TSs showing endogenous loss at 13q12.3 at D13S1299, whereas both KYSE140 and KYSE180 cell lines contain D13S1229, and D13S1226; 13q13.1 at D13S260; and 13q13.2

at D13S267 are colocalized with the nonhomozygosity regions did not support a role for BRCA2 in SLMT-1 tumorigenesis. of chromosome 13q in SLMT-1 at 13q12.3-13.3 (Fig. 1). Unknown TSG(s) within 13q12.3 remain to be identified. The reason for favoring endogenous loss at 13q12.3-13q13.2 is unknown and may be related to a gene dosage effect (13). CRs at 13q14: Novel Candidate TSG Other than RB1 Of interest, a 0.373 Mb CR2, mapped to 13q12.3, was RB1 is a well-known TSG mapped to 13q14, but no studies colocalized to the same CR identified in a NPC model system of RB1 cDNA transfer in ESCC and mutation of RB1 in in a chromosome 13 transfer study (14). CR1 and CR2 are both primary ESCC have been reported. In contrast to the high LOH located at 13q12. LOH at 13q12 occurs frequently in ESCC reported at the RB1 locus (16), RB1 protein expression was (6-8) and is associated with lymph node metastasis (15). absent in only a small subset (6.4%, 11 of 172) of ESCC BRCA2 is a candidate TSG at 13q12-13, but both the micro- patients by immunohistochemical staining (17). RB1 is outside array gene expression profile analysis and TSdeletion analysis the 13q14 LOH region mapped by Li et al. (9). Thus, these

FIGURE3. A. RT-PCR of eight candidate TSGs (KIAA0853, ESD, CHC1L, PHF11, RFP2, FLJ11712, THSD1, and C13ORF9) identified by microarray. They all show up-regulation in three chromosome 13 MCHs (MCH13-111, MCH13-113, MCH13-117) and the immortalized NE1 cell line, compared with the recipient SLMT-1 cell line. They are down-regulated in the other 15 ESCC, A549, and HeLa cell lines, compared with NE1. All 18 cancer cell lines showed THSD1 and PHF11 down-regulation. B. Real-time quantitative RT-PCR analysis of THSD1 and PHF11 in 16 ESCC cell lines, HeLa, and A549 versus NE1. The percentage of cell lines showing at least 10-fold down-regulation of THSD1 and PHF11 were 77.8% (14 of 18) and 44.4% (8 of 18), respectively.

FIGURE4. A. Transient transfection of SLMT-1 with pCR3.1-THSD1 v1, pCR3.1-THSD1 v2, or pCR3.1. RT-PCR shows the up-regulation of THSD1 1d after transfection. B. Colony formation assay of THSD1 transfectants. SLMT-1 was transfected with pCR3.1, pCR3.1-THSD1 v1, and pCR3.1-THSD1 v2. Representative results from four independent experiments showed a significant reduction of colonies after THSD1 overexpression.***, P < 0.001, a significant difference in colony numbers from vector alone. C. MSP analysis of THSD1 promoter methylation status in six ESCC cell lines and NE1. Both the methylated (M) and unmethylated (U) alleles were observed in all ESCC cell lines. Only the unmethylated allele was observed in NE1. D. Map of the 338-bp CpG island around exon 1 in THSD1. Vertical bars, CpG sites. Results of bisulfite sequencing done in the THSD1-expressing NE1 cell line and THSD1 down-regulated KYSE140 and KYSE180 ESCC cell lines, and two pairs (2N/2T and 80N/80T) of THSD1 down-regulated matched nontumor and ESCC tumor tissues. Ten clones were sequenced in each cell line and tissue. The percentage of methylation in each CpG site is denoted by pie charts, as indicated. E. Demethylation treatment with 5-aza-2¶-deoxycytidine (5-Aza-dC) in both EC18 and KYSE180 resulted in increase of THSD1 expression, as monitored by RT-PCR. TSA, trichostatin A. F. Quantitative RT-PCR showed a synergistic effect of THSD1 reexpression in both cell lines, reaching the level of NE1, when cotreated with 5-Aza-dC and trichostatin A. studies suggest that unknown novel TSG(s), other than RB1, 1.89-fold increase in RB1 expression (Table 1). Western blot are responsible for the high allelic loss at 13q14 and remain to analysis of the three MCHs showed slight increases in total RB be identified for ESCC tumorigenesis. protein. However, when compared with three hTERT immor- The present study used the SLMT-1 cell line, which has a talized esophageal epithelial cell lines, the normal active form reduced level of the 110 kDa RB1 protein expression of RB is not increased in the MCHs (Supplementary Fig. S2). (Supplementary Fig. S2). From the microarray analysis, the In addition to RB1 protein expression analysis in SLMT-1, the three tumor-suppressing MCH cell lines showed an average of RB1 marker, D13S153, located at exon 2 of RB1, was not

deleted in any of the eight TSs. The complete loss of p16INK4a addition to partial methylation, THSD1 inactivation in 80N/T in SLMT-1 (18), together with the current deletion analysis of likely involves LOH, because LOH was detected in the DNA TSs by PCR-based microsatellite typing and RB1 Western from the same pair of tissues (80N/T) using two microsatellite blotting of chromosome 13 MCHs, provide evidence that RB1 is markers in close proximity to THSD1 (D13S284 and D13S133, not the target TSG at 13q14 responsible for the tumor suppres- data not shown). Although 10% to 30% methylation was sion observed in SLMT-1, the current ESCC model system. observed with the 2N specimen obtained from the nontumorous margin of the esophageal tissue resection, the tumor tissue 2T Novel Candidate TSG at 13q14: THSD1 showed an increased methylation intensity of 20% to 60%. The current functional studies implicate the importance of Previous investigators have also reported a lower extent of chromosome 13q14 in ESCC tumor suppression. TS micro- methylation of nontumorous tissues compared with tumor satellite deletion mapping analysis localized two CRs (CR3 at specimens (19). The results of this study indicate that the D13S263 and CR4 at D13S133) at chromosomal region 13q14, THSD1 promoter is partially and heterogeneously hyper- which are frequently eliminated. Microarray analysis of methylated. Further studies are required to determine whether differential gene expression identified nine 13q14 candidates full restoration of THSD1 expression to the immortalized genes up-regulated in the three tumor-suppressing chromosome normal epithelial NE1 levels after trichostatin A treatment can 13 MCHs. RT-PCR was done for KIAA0853, ESD, RB1, be attributed to other epigenetic events such as histone CHC1L, PHF11, RFP2, FLJ11712, THSD1, and C13orf9. deacetylation. Such a synergism between 5-aza-2¶-deoxycyti- Results validated the frequent down-regulation of THSD1 and dine and trichostatin A has been shown for many genes. PHF11 in a panel of 18 cancer cell lines. THSD1 is located Methylated promoters may recruit methyl CpG-binding pro- between FLJ11712 and C13orf9 (Fig. 1) within 13q14.3, but teins (MeCP2), which, in turn, recruit histone deacetylases and only specific loss of THSD1 expression in all cancer cell lines other corepressors, to form large protein complexes. The was detected. Because THSD1 showed a more prominent loss binding of the large protein complexes with DNA leads to an than that of PHF11, it was chosen for further functional inactive chromatin structure and blocks gene transcription (20). characterization in this current study. PHF11 remains another Hence, the maximal reactivation of genes silenced by interesting target for further investigations in a future study. The methylation may require simultaneous blockage of DNA wild-type THSD1 transfection in SLMT-1 resulted in signifi- methylation and histone deacetylation. cant reduction of colony formation ability, providing evidence The function of THSD1 is unknown. It encodes a for a growth-suppressive role of THSD1 in ESCC tumorigen- transmembrane molecule containing a thrombospondin type 1 esis. In six ESCC cell lines showing 50-fold to 600-fold down- repeat, which may be involved in cell adhesion and regulated THSD1 expression compared with NE1, both angiogenesis. A review on thrombospondins and tumor methylated and unmethylated alleles were observed (Fig. 4C). angiogenesis commented that proteins possessing the thrombo- Additionally, demethylation treatment increased THSD1 ex- spondin type 1 repeat, including thrombospondin-1, ADAM pression in EC18 and KYSE180 (Fig. 4E and F), further metallopeptidase with thrombospondin type 1 motif 1 indicating that hypermethylation is involved in THSD1 down- (ADAMTS1) and ADAMTS8, and brain-specific angiogenesis regulation. Promoter methylation may be one of the mecha- inhibitor 1 (BAI1), regulate angiogenesis (21). Recently, Myc- nisms responsible, at least in part, for the loss of THSD1 induced enhanced neovascularization in mouse colonocytes expression in ESCC tumorigenesis. For KYSE140 and 2T, this correlated with down-regulation of antiangiogenic thrombo- seems to be an important inactivation mechanism. For spondin-1 and other proteins with thrombospondin type 1 KYSE180 and 80T, which do not show such extensive repeat, such as connective tissue growth factor and THSD1 methylation, both epigenetic silencing and LOH are likely (22), suggested that THSD1 may be involved in tumor involved in the inactivation of THSD1. Clonal bisulfite angiogenesis. An interaction network for THSD1 v1 predicted 9 sequencing results with cell lines (Fig. 4D) indicate that the by ‘‘SMART’’ includes vascular cell adhesion molecule 1, normal NE1 does not contain methylated sequences, whereas intercellular adhesion molecule 1 (CD54, human rhinovirus KYSE180 contains 10% to 50% partial methylated sequences receptor), intercellular adhesion molecule 2, and mucosal and KYSE140 contains more extensive, 80% to 100%, vascular addressin cell adhesion molecule 1. Interestingly, methylated sequences. The partial and extensive methylation analysis of the differential expression levels of this gene in in KYSE180 and KYSE140, respectively, detected by bisulfite previous microarray studies show that high THSD1 expression sequencing (Fig. 4D), were concordant with the observation of positively correlated with a better distant metastasis survival in a much more intense band of the M allele detected by MSP breast cancer patients (Supplementary Fig. S3). This is analysis for KYSE140 versus KYSE180 (Fig. 4C). The consistent with its loss possibly being associated with presence of both the methylated and unmethylated alleles in metastatic tumor spread; studies are needed to evaluate its KYSE 180 detected in MSP analysis and partial methylated potential importance as a biomarker for esophageal carcinoma. CpG sites observed by bisulfite sequencing reflect heteroge- Further functional studies on THSD1 are now under way to neous methylation or a mix of cells carrying methylated or elucidate its tumor-suppressive role. unmethylated alleles, with one allele lost by LOH (Fig. 1). Differential gene expression profiles of three tumor- Comparative examination of tumor tissues and their adjacent suppressing MCHs and their tumorigenic parental SLMT-1 cell nontumor margins showed greater hypermethylation in tumors. None of the CpG sites were methylated in the 80N specimen, but its tumor tissue showed 10% to 30% partial methylation. In 9 http://smart.embl-heidelberg.de/

line were revealed by cDNA oligonucleotide microarrays. A list (MCH13-111, MCH13-113, MCH13-117) and two TScell lines of 40 genes (Table 1B), showing down-regulation in SLMT-1 (MCH13-111/TS2R and MCH13-117/TS1L). Eight TS cell compared with NE1, the immortalized normal esophageal lines were established. Details of the tumorigenicity assay and epithelial cell line, in regions outside the critical 13q regions criteria for TSdeletion mapping analyses were previously identified on chromosome 13q, might be involved in global described (12). downstream cascades of tumor suppression in the nude mouse model. The identified genes had a wide spectrum of functions, RNA Extraction and Oligonucleotide Microarray Hybrid- including cell proliferation, cell death regulator, transcription ization factor, signal transduction, extracellular matrix, cell adhesion, Total RNA was extracted from the cell lines with RNeasy immune response, and heat shock proteins. Of these, CDKN1A Midi Kit (Qiagen). Twenty micrograms of total RNA were is a cyclin-dependent kinase inhibitor that negatively regulates reverse transcribed with SuperScript II reverse transcriptase cell proliferation. Their importance in ESCC tumorigenesis and (Invitrogen) into cDNA. The cDNAs of SLMT-1 and MCHs usefulness as biomarkers remains to be further evaluated. and NE1 were labeled with Cy5-dUTP and Cy3-dUTP (Amersham Biosciences), respectively. Reciprocal dye swap Materials and Methods labeling was done in each case. For SLMT-1/NE1 comparison, Cell Lines and Culture Conditions the whole experiment was repeated twice. The high-density Human ESCC recipient cell line, SLMT-1, and mouse donor microarrays used for our experiments were spotted with 18,912 cell line containing the intact human chromosome 13, 60-mer oligonucleotides from a human oligonucleotide library MCH204.3, were used for microcell-mediated chromosome (Sigma-Genosys) onto poly-L-lysine–coated glass slides at the transfer. Culture conditions for the SLMT-1, donor, MCH, and Genome Institute of Singapore (24). The labeled probes were TScell lines, and details for a panel of 16 ESCCcell lines were hybridized to the 19k chips in a Maui hybridization chamber described previously (10, 12). NE1, a human papillomavirus (BioMicro Systems) for 16 h at 42jC. The slides were washed E6/E7/telomerase–immortalized normal esophageal epithelial sequentially in 2Â SSC and 0.1% SDS for 1 min, 1Â SSC for cell line (23), is the reference cell line for gene expression 30 s, 0.2Â SSC for 30 s, 0.05Â SSC for 8 s, and then spin profile analysis in the microarray and RT-PCR studies. dried. Hybridized arrays were scanned on a GenePix 4100A scanner (Axon Instruments) at 635 and 532 nm for Cy5- and Microcell-Mediated Chromosome Transfer Cy3-labeled cDNAs. The images were analyzed by GenePix The intact chromosome 13 was introduced into the ESCC Pro 4.0 and the resulting data was processed by the mAdb SLMT-1 cell line using microcell-mediated chromosome microarray database of the Genome Institute of Singapore. transfer techniques, as described previously (12). Chromosome Candidates were shortlisted, when intensities of hybridization z 13 MCHs were obtained after selection for 4 weeks. signals were 2 in four of four slides for SLMT-1/NE1 and z1.2 in at least five of six slides in the three cases of SLMT/ DNA Extraction, Slot Blot Hybridization, and FISH chromosome 13 MCHs. Genomic DNAs from the donor, recipient, MCHs, and TSs were extracted, as previously described (12). The MCHs were Semiquantitative and Real-time RT-PCR tested for the absence of mouse DNA contamination by DNA For RT-PCR, total RNAs were reverse transcribed with slot blot hybridization, as previously described (12). The Moloney murine leukemia virus (GE Healthcare) into cDNA, presence of the extra chromosome 13 in the MCHs was verified and PCR was done with AmpliTaq Gold DNA polymerase by FISH using WCP 13 SpectrumGreen chromosome 13– (Applied Biosystems) as previously described (11). Details of specific probes (Vysis). FISH and image capture were done, as primer sequences are shown in Table 2. Real-time quantitative previously described (11). A minimum of 20 metaphase spreads PCR was done using Mx3000P real-time PCR system was analyzed. (Stratagene; ref. 10). The THSD1 (transcript variant 1), PHF11, and GAPDH Taqman probes were purchased from PCR Microsatellite Typing Assay Applied Biosystems. Fifty microsatellite markers spanning the entire chromosome 13 were used. PCR amplification and capillary electrophoresis MSP and Bisulfite Sequencing Analysis of PCR products by the semiautomated ABI PRISM 3100 A fragment of 910 bp at location À808 to +101 genetic analyzer were done as described (12). The mapping (chr13:51878221-51879130) was identified as the putative information (Fig. 1) and primer sequences were obtained from THSD1 promoter by Gene2Promoter. One CpG island of 338 the University of California Santa Cruz Genome Bioinfor- bp was located within the putative promoter of THSD1 with an matics10 and National Center for Biotechnology Information observed/expected CpG ratio, 0.60; a total of 35 CpG sites were genome databases.8 found. THSD1 methylation status in NE1 and ESCC cell lines was determined by chemical treatment with sodium metabi- Tumorigenicity Assay and TS Analyses sulfite, MSP analysis, and bisulfite sequencing. Briefly, 2 Ag Each of the six injection sites was inoculated with 2 Â 106 genomic DNA were modified by sodium metabisulfite at 55jC cells in the nude mouse assay for three chromosome 13 MCHs for 15 h. Bisulfite-treated DNA was purified with QIAquick PCR Purification Kit according to the manufacturer’s protocol and was ethanol precipitated. Primer sequences for MSP 10 http://genome.ucsc.edu/ analysis and bisulfite sequencing (Table 2) were designed by

TABLE2. Details of RT-PCR and THSD1 MSP Analysis and Bisulfite Sequencing Primer Sequences

Gene Nucleotide Positions Size (bp) Primer Sequences

KIAA0853 1539-1559 501 5¶-AAAGAGAGCCGTGATCCCAGA-3¶ 2039-2019 5¶-ACGCTCAAGCTCGTCATTCCT-3¶ ESD 434-454 295 5¶-GCCCTCGTGGCTGCAATATTA-3¶ 748-728 5¶-TAAAGGCTTTTTTGCCCCAGG-3¶ CHC1L 278-298 401 5¶-GGAGCTGAGAATTGATGTCCG-3¶ 678-659 5¶-GGCAGGCTATTTTTTTGCCA-3¶ PHF11 132-152 552 5¶-ACCGGTGTCTTTCAGGTTGCA-3¶ 683-663 5¶-TGCATCTGTGTGTCTGCCATG-3¶ RFP2 583-604 309 5¶-GGTGCTGTGCAAATAAATGCGT-3¶ 891-869 5¶-GGAGCTGGTCTCCACAAGGAAT-3¶ FLJ11712 422-444 251 5¶-TGTTCAGGAGAAGGAGCCATTT-3¶ 672-650 5¶-AAGATGCAATTTGGAAACACGTT-3¶ THSD1 773-794, 1652-1631 (transcript variant 1) 880 5¶-AAGAAATTCAAGACAGCCGCTG-3¶ 5¶-GCTCAGCTCGCAGATATTCTCC-3¶ 715-736, 1435-1414 (transcript variant 2) 721

C13orf9 746-765 401 5¶-CAGTACCACATGCAGCTGGC-3¶ 1146-1127 5¶-TCCACTGAGTCATCACGGCA-3¶ GAPDH 58-76 226 5¶-GAAGGTGAAGGTCGGAGTC-3¶ 283-264 5¶-GAAGATGGTGATGGGATTTC-3¶ Methylated THSD1 (À69)-(À90) 217 5¶-GTTTTCGTTTTTTTTGGGTAGC-3¶ (À285)-(260) 5¶-AAAATATCAAAAAATTACTCGCTCGT-3¶ Unmethylated THSD1 (À72)-(À93) 209 5¶-TTTGTTTTTTTTGGGTAGTGTA-3¶ (À280)-(À258) 5¶-ATCAAAAAATTACTCACTCATCC-3¶ THSD1-L1 (À121)-(À146) 211 5¶-GTTGTTGGATAGGAAATAGGTAGGA-3¶ THSD1-R1 (À331)-(À301) 5¶-AATCTCCCTAAAAAAAATAACACTAAAATA-3¶

MethPrimer v1.1 h (25). Bisulfite-treated genomic DNAs were Statistical Analysis amplified by PCR, gel purified, and subcloned into pMD18-T The independent t test was used for nude mouse assay and vectors (TaKaRa Biotech). Ten clones from each cell line colony forming assay. or tissue were sequenced. The PCR amplicon encompassed 12 CpG sites in the À121 to À331 region of the CpG island around THSD1 exon 1. References 1. Pisani P, Parkin DM, Bray F, Ferlay J. Estimates of the worldwide mortality from 25 cancers in 1990. Int J Cancer 1999;83:18 – 29. 5-Aza-2¶-Deoxycytidine and Trichostatin A Treatment 2. Hu N, Roth MJ, Polymeropolous M, et al. Identification of novel regions of EC18 and KYSE180 were treated with 5 Amol/L 5-aza-2¶- allelic loss from a genomewide scan of esophageal squamous-cell carcinoma in a high-risk Chinese population. Genes Chromosomes Cancer 2000;27:217 – 28. deoxycytidine (Sigma), as previously described (10). On the 5th 3. Pack SD, Karkera JD, Zhuang Z, et al. Molecular cytogenetic fingerprinting of day of 5-aza-2¶-deoxycytidine treatment, 300 nmol/L trichos- esophageal squamous cell carcinoma by comparative genomic hybridization tatin A (Sigma) was added for 24 h before cell harvesting. reveals a consistent pattern of chromosomal alterations. Genes Chromosomes Cancer 1999;25:160 – 8. 4. Tsang YS, Lo KW, Leung SF, et al. Two distinct regions of deletion on Cloning of THSD1, Gene Transfection, and Colony chromosome 13q in primary nasopharyngeal carcinoma. Int J Cancer 1999;83: Formation Assay 305 – 8. The full-length wild-type cDNAs of both THSD1 variants 5. Tamura K, Zhang X, Murakami Y, et al. Deletion of three distinct regions on chromosome 13q in human non-small-cell lung cancer. Int J Cancer 1997;74: were cloned from NE2, an immortalized epithelial cell line, 45 – 9. ¶ using the following primers: THSD1-NheI-start-2 5 -TACGGC- 6. Hu N, Goldstein AM, Albert PS, et al. Evidence for a familial esophageal TAGCATGAAACCAATGTTGAAAGAC-3¶ (forward) and cancer susceptibility gene on chromosome 13. Cancer Epidemiol Biomarkers THSD1-Stop-NotI-2 5¶-ATAGTTTAGCGGCCGCCTAGAT- Prev 2003;12:1112 – 5. CACCAGCTTCTCC-3¶ (reverse). The complete coding se- 7. Huang XP, Wei F, Liu XY, et al. Allelic loss on 13q in esophageal squamous cell carcinomas from northern China. Cancer Lett 2002;185:87 – 94. quence of THSD1 v1 and v2 contains 2,559 and 2,400 bp, 8. Li G, Hu N, Goldstein AM, et al. Allelic loss on chromosome bands 13q11-13 respectively. The coding sequence of THSD1 v1 and v2 was in esophageal squamous cell carcinoma. Genes Chromosomes Cancer 2001;31: sequence-verified with reference to the Genbank accession no. 390 – 7. NM_018676 (GI:40805850) and NM_199263 (GI:40805851), 9. Li WJ, Hu N, Su H, et al. Allelic loss on chromosome 13q14 and mutation in respectively. Subsequently, both variants were cloned into deleted in cancer 1 gene in esophageal squamous cell carcinoma. Oncogene 2003; 22:314 – 8. pCR3.1 neomycin-resistant plasmid with NheIandNotI. 10. Lo PH, Leung AC, Kwok CY, et al. Identification of a tumor suppressive One microgram of plasmid DNA was transfected into critical region mapping to 3p14.2 in esophageal squamous cell carcinoma and SLMT-1 with Lipofectamine 2000 reagent (Invitrogen) for studies of a candidate tumor suppressor gene, ADAMTS9. Oncogene 2007;26: 4.5 h. After 4 weeks of selection in DMEM/5% FCScontaining 148 – 57. A 11. Yang L, Leung AC, Ko JM, et al. Tumor suppressive role of a 2.4 Mb 9q33- 500 g/mL neomycin, colonies were fixed and stained with 34 critical region and DEC1 in esophageal squamous cell carcinoma. Oncogene Giemsa for colony scoring. The assays were repeated inde- 2005;24:697 – 705. pendently in quadruplicate. 12. Ko JM, Yau WL, Chan PL, et al. Functional evidence of decreased

tumorigenicity associated with monochromosome transfer of chromosome 14 in 19. Zhong S, Yeo W, Tang MW, Wong N, Lai PB, Johnson PJ. Intensive esophageal cancer and the mapping of tumor-suppressive regions to 14q32. Genes hypermethylation of the CpG island of Ras association domain family 1A in Chromosomes Cancer 2005;43:284 – 93. hepatitis B virus-associated hepatocellular carcinomas. Clin Cancer Res 2003;9: 13. Kost-Alimova M, Darai-Ramqvist E, Yau WL, et al. Mandatory chromo- 3376 – 82. somal segment balance in aneuploid tumor cells. BMC Cancer 2007;7:21. 20. Kaludov NK, Wolffe AP. MeCP2 driven transcriptional repression in vitro: 14. Cheng Y, Lung HL, Wong PS, et al. Chromosome 13q12 region critical for selectivity for methylated DNA, action at a distance and contacts with the basal the viability and growth of nasopharyngeal carcinoma hybrids. Int J Cancer 2004; transcription machinery. Nucleic Acids Res 2000;28:1921 – 8. 109:357 – 62. 21. de Fraipont F, Nicholson AC, Feige JJ, Van Meir EG. Thrombospondins and tumor angiogenesis. Trends Mol Med 2001;7:401 – 7. 15. Harada H, Tanaka H, Shimada Y, Shinoda M, Imamura M, Ishizaki K. Lymph node metastasis is associated with allelic loss on chromosome 13q12-13 in 22. Dews M, Homayouni A, Yu D, et al. Augmentation of tumor angiogenesis by esophageal squamous cell carcinoma. Cancer Res 1999;59:3724 – 9. a Myc-activated microRNA cluster. Nat Genet 2006;38:1060 – 5. 16. Boynton RF, Huang Y, Blount PL, et al. Frequent loss of heterozygosity at 23. Deng W, Tsao SW, Guan XY, et al. Distinct profiles of critically short the retinoblastoma locus in human esophageal cancers. Cancer Res 1991;51: telomeres are a key determinant of different chromosome aberrations in 5766 – 9. immortalized human cells: whole-genome evidence from multiple cell lines. 17. zur Hausen A, Sarbia M, Heep H, Willers R, Gabbert HE. Retinoblastoma- Oncogene 2004;23:9090 – 101. protein (prb) expression and prognosis in squamous-cell carcinomas of the 24. Lung HL, Bangarusamy DK, Xie D, et al. THY1 is a candidate tumour esophagus. Int J Cancer 1999;84:618 – 22. suppressor gene with decreased expression in metastatic nasopharyngeal 18. Kwong FM, Tang JC, Srivastava G, Lung ML. Inactivation mechanisms and carcinoma. Oncogene 2005;24:6525 – 32. growth suppressive effects of p16INK4a in Asian esophageal squamous 25. Li LC, Dahiya R. MethPrimer: designing primers for methylation PCRs. carcinoma cell lines. Cancer Lett 2004;208:207 – 13. Bioinformatics 2002;18:1427 – 31.

Mol Cancer Res 2008;6(4). April 2008 Downloaded from mcr.aacrjournals.org on September 26, 2021. © 2008 American Association for Cancer Research. Monochromosome Transfer and Microarray Analysis Identify a Critical Tumor-Suppressive Region Mapping to Chromosome 13q14 and THSD1 in Esophageal Carcinoma

Josephine M.Y. Ko, Pui Ling Chan, Wing Lung Yau, et al.

Mol Cancer Res 2008;6:592-603.

Updated version Access the most recent version of this article at: http://mcr.aacrjournals.org/content/6/4/592

Supplementary Access the most recent supplemental material at: Material http://mcr.aacrjournals.org/content/suppl/2008/06/17/6.4.592.DC1

Cited articles This article cites 25 articles, 4 of which you can access for free at: http://mcr.aacrjournals.org/content/6/4/592.full#ref-list-1

Citing articles This article has been cited by 1 HighWire-hosted articles. Access the articles at: http://mcr.aacrjournals.org/content/6/4/592.full#related-urls

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

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

Permissions To request permission to re-use all or part of this article, use this link http://mcr.aacrjournals.org/content/6/4/592. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.