Integrative Genomics Identifies RAB23 As an Invasion Mediator Gene in Diffuse-Type Gastric Cancer

Research Article

Integrative Genomics Identifies RAB23 as an Invasion Mediator Gene in Diffuse-Type Gastric Cancer

Qingsong Hou,1 Yong Hui Wu,1 Heike Grabsch,5 Yansong Zhu,1 Siew Hong Leong,1 Kumaresan Ganesan,1 Debra Cross,5 Lay Keng Tan,3 Jiong Tao,1 Veena Gopalakrishnan,1 Bor Luen Tang,2 Oi Lian Kon,1 and Patrick Tan1,3,4

1National Cancer Centre, 2Department of Biochemistry, National University of Singapore, 3Duke-NUS Graduate Medical School, and 4Genome Institute of Singapore, Singapore, Singapore; and 5Leeds Institute of Molecular Medicine, St James’s University Hospital, Leeds, United Kingdom

Abstract genes. For example, chromosome 6p amplifications have been reportedin GC at frequencies ranging from 2% to 4% by Recurrent genomic amplifications and deletions are frequently observed in primary gastric cancers (GC). However, identifying comparative genomic hybridization (CGH) to 85% by cDNA-based specific oncogenes and tumor suppressor genes within these array-basedcomparative genomic hybridization(aCGH; refs. 2, 8, 9), regions can be challenging, as they often cover tens to hundreds but specific genes representing the target of 6p amplification are of genes. Here, we combined high-resolution array-based currently unknown. comparative genomic hybridization (aCGH) with gene expres- The importance of chromosomal aberrations in GC is also sion profiling to target genes within focal high-level amplifica- highlightedby the underlyinggenetics of its histologic subtypes. tions in GC cell lines, and identified RAB23 as an amplified and GCs are broadly classified into intestinal (iGC) or diffuse (dGC) overexpressed Chr 6p11p12 gene in Hs746T cells. High RAB23 forms (10), which exhibit strikingly different morphologies. iGCs protein expression was also observed in some lines lacking form cohesive gland–like tubular structures, whereas dGCs exhibit RAB23 amplification, suggesting additional mechanisms for minimal cell cohesion with frequent invasion into surrounding up-regulating RAB23 besides gene amplification. siRNA silenc- tissues (11). Although amplifications at 17q12-21 and20q and ing of RAB23 significantly reduced cellular invasion and losses at 8p and9p are more frequently observedin iGC, 8p and 12q gains are associatedwith dGC(12, 13), arguing that iGC migration in Hs746T cells, whereas overexpression of RAB23 RAB23 and dGC may represent molecularly distinct entities. Furthermore, enhanced cellular invasion in AGS cells. amplifications ERBB2/HER2 in primary gastric tumors were confirmed by both fluorescence amplifications are observedin iGC but not dGCs, in situ hybridization and genomic qPCR, and in two indepen- raising the possibility that only the former may be candidates for dent patient cohorts from Hong Kong and the United Kingdom Trastuzumab (anti-HER2) therapy (14). Identifying additional RAB23 expression was significantly associated with diffuse-type genetic andgenomic aberrations specifically associatedwith iGC GC (dGC) compared with intestinal-type GC (iGC). These results or dGC might further elucidate the mechanistic basis of their provide further evidence that dGC and iGC likely represent two morphologic differences, and identify potential avenues for iGC or molecularly distinct tumor types, and show that investigating dGC-specific treatments. focal chromosomal amplifications by combining high- Recently, high-resolution aCGH platforms have alloweddetec- tion of genomic aberrations that are <40 Kb (15). In this report, we resolution aCGH with expression profiling is a powerful strategy for identifying novel cancer genes in regions of usedhigh-resolution aCGH to analyze a panel of GC cell lines and recurrent chromosomal aberration. [Cancer Res 2008;68(12): discovered a focal high level amplification at Chr 6p11p12 in 4623–30] Hs746T cells that has not been previously described. We identified RAB23 as an amplifiedandoverexpressedgene locatedin this region, andconfirmed RAB23 amplifications in primary tumors. We Introduction provide functional evidence that RAB23 regulates cell invasion in Chromosomal instability is frequently observedin gastric cancer GC, and in two independent patient cohorts, RAB23 overexpression (GC), with up to 90% of primary tumors exhibiting aneuploidy (1). at both the RNA andprotein level was strongly associatedwith Previous studies have revealed a complex portrait of recurring dGC, providing further evidence that dGCs and iGCs likely develop chromosomal amplifications and deletions in GC, including gains via different molecular pathways (16, 17). Taken collectively, our andlosses at chromosomes 1q, 8q, 17q, and20q (2–4). The recurrent study shows that targeting focal chromosomal amplifications using nature of these aberrations has been attributedto the presence of several different high-resolution techniques can lead to the rapid genes important for gastric carcinogenesis, such as CD44 at 11p13, identification of candidate genes relevant to GC. Such strategies CCNE1 at 19q13, and BTAK at 20q13 (5–7). However, identifying might also be extended to other cancer types. specific oncogenes andtumor suppressor genes within such regions is often challengedby their large sizes covering tens to hundredsof Materials and Methods Cell culture. GC cell lines were purchasedfrom the American Type Note: Supplementary data for this article are available at Cancer Research Online Culture Collection andculturedin 90% DMEM (high glucose) + 10% fetal (http://cancerres.aacrjournals.org/). bovine serum (FBS) for Hs746T; 90% RPMI + 10% FBS for AGS, NCI-N87, Requests for reprints: Patrick Tan, Duke-NUS Graduate Medical School, 2 Jalan SNU-1, SNU-5, andSNU-16; and80% RPMI + 20% FBS for KATOIII at 5% Bukit Merah, Singapore, 169547 Singapore. Phone: 65-6-436-8345; Fax: 65-6-226-5694; CO ,37jC. E-mail: [email protected]. 2 I2008 American Association for Cancer Research. BAC array-CGH. Genomic DNA was extractedusing a Qiagen extraction doi:10.1158/0008-5472.CAN-07-5870 kit. Two micrograms of cell line andreference genomic DNA (unrelated www.aacrjournals.org 4623 Cancer Res 2008; 68: (12). June 15, 2008

Downloaded from cancerres.aacrjournals.org on October 3, 2021. © 2008 American Association for Cancer Research. Cancer Research pooledmale lymophocyte DNA from healthy donors) were labeledwith siRNA and cDNA transfection. Hs746T andAGS cells were seeded Cy3-dCTP or Cy5-dCTP, respectively, (Amersham Biosciences) and hybrid- into a 24-well plate at a density of 1.8 Â 105 cells/mL andincubatedfor izedonto microarrays containing 32,000 DOP-PCR–amplifiedBAC clones 24 h before transfection with either negative control siRNA or specific covering the whole human genome (Children’s Hospital Oakland Research RAB23 siRNAs (200 nmol/L, SMART pool or individual RAB23 siRNAs; Institute, BACPAC resources; ref. 15). Hybridizations were performed Dharmacon) by Oligofectamine (Invitrogen) in OptiMEM Medium (Invi- using a MAUI hybridization station, and scanned using a GenePix 4000B trogen). The ‘‘negative control’’ siRNA was a scrambledsiRNA containing a scanner (Axon Instruments). Raw images were analyzedusing GenePix Pro random nucleotide sequence. For RAB23 overexpression, a full-length 4.0 software. RAB23 cDNA was clonedinto the pCI-Neo mammalian expression vector 100K single-nucleotide polymorphism array. Genomic DNAs were (Promega) andtransfectedinto AGS cells using Oligofectamine reagent hybridized onto Affymetrix GeneChip Human Mapping 50K XbaI arrays and (Invitrogen). After 72 h of siRNA treatment, relative cell proliferation was 50K HindIII arrays according to the manufacturer’s instructions. Raw image determined by measuring absorbance using the 3-(4,5-dimethylthiazol-2-yl)- files were analyzedusing Dchip Software. 6 The microarray data in this study 2,5-diphenyltetrazolium bromide assay (Roche). Each assay was carried out has been deposited in the Gene Expression Omnibus database under in triplicate, andnormalizedrelative to control-transfectedcells. accession number GSE10611. Invasion and migration assays. Cell invasion assays were performed Gene expression profiling. RNA was extractedusing Trizol using Biocoat Matrigel invasion chambers with 8-Am pore filter inserts reagent (Invitrogen) andprocessedfor Affymetrix Genechip hybridizations (BD Bioscience). The 8-Am pore size was selectedfor these experiments as on U133A Genechips according to the manufacturer’s instructions. this size has been usedin other publishedstudies(19–21). Forty-eight hours Gene expression data were quality controlled by GeneData Refiner.7 Gene after transfection, Hs746T or AGS cells were trypsinizedandtransferred expression data from individual arrays were normalized by median to the upper Matrigel chamber in 500 AL of serum-free medium containing centering around500 expression units. Two or three experimental replicates 1 Â 105 cells (5 Â 104 for AGS cells) andincubatedfor 44 h. The 44-h time were performedfor each line (18). point was selectedbasedon the intrinsic invasive rate of Hs746T cells, Fluorescence in situ hybridization analysis. Single-cell nuclear which is different between different cell lines. Migrating cells were counted suspensions were preparedusing standardinterphasefluorescence in situ using light microscopy. In the figures, we have insertedredarrows to hybridisation analysis (FISH) protocols. BACs specific to 6p11p12 (RP11- highlight representative cells invading through the transwell, which should 641O11; CHORI) were labeled with Texas red–conjugated dUTP and mixed be distinguished from the 8-Am pores of the transwell material (small round with FITC-labeledcentromeric 6 probes (D6Z1; American Type Culture circles). Each assay was performedin triplicate, andthe results were Collection). Up to 100 interphase nuclei andmetaphases were assessed,and averaged over three independent experiments. For cell migration assays, fluorescent images were analyzedusing FISHview (AppliedSpectral Hs746T cells were transfectedwith control or RAB23 siRNAs andseeded Imaging; GmbH). For primary GCs, imprints from 10 primary GCs and into a 6-well plate to reach 100% confluency at the time of maximum RAB23 matchednormal gastric mucosa from fresh resection specimens were knockdown. A ‘‘scratch’’ wound was inflicted across each well using a placedon glass slides.Probes for the primary tumor experiments were 200-AL pipette tip, andthe rate of cells migrating into the scratch was labeledwith spectrum orange (D6Z1) or FITC (RP11-641O11). observed. Pictures of the wound were taken at 24- and 48-h intervals, and Real-time genomic PCR. Quantitative real-time PCR was performedon cell migration distances across the wound were measured. Six parallel a LightCycler (Roche) using SYBR Green PCR Master Mix (Roche) and experiments were performedin a total of 24 wells. primers specific to the RAB23 and h-ACTIN (ACTB) genomic locus: Clinical samples. We testedthree differentpatients cohorts, andthe RAB23_fwd, 5¶-AGCGAGACTCCGTCTTCAAA; RAB23_rev, 5¶-CACCCC- type of experiment performedon each cohort was basedon the availability TAAGGTACGCATGT; ACTB_fwd, 5-¶GTGGCATCCACGAAACTACC; ACT- of genomic DNA, gene expression information, clinical data, or tissue B_rev, 5-¶AGCTCAGGCAGGAAAGACAC. Quantification of genomic copy microarrays (TMA). number was basedon standardcurves, andeach qPCR was performed Singapore: Anonymizedprimary human gastric tumors were obtained either in duplicate or triplicate. For both cell lines and primary tumors, from the National Cancer Centre of Singapore tissue repository with RAB23 gene copy numbers were comparedrelative to the ACTB genomic approval from the local Ethics Committee andsignedpatient informed control within the same sample. For the cell line analysis, normal diploid consent. Tumor content in all samples was confirmedto be >50% by frozen DNAs were not included because the main purpose of this experiment was sections. Genomic DNA from pooledhealthy female andmale blood to compare RAB23 gene copy number levels across the lines. In contrast, lymphocytes were usedas normalization controls. normal diploid DNAs were used in the analysis of primary tumor samples to Hong Kong: Gene expression data for 90 GC patients and 28 normal show increased RAB23 gene copy number levels in tumors comparedwith gastric tissues were obtained from genome-www.stanford.edu.8 normal tissues. United Kingdom (Leeds): 306 patients receiving surgical resections for Semiquantitative reverse transcription-PCR. RNA was reverse tran- gastric adenocarcinoma at the Academic Department of Surgery, Leeds scribedinto cDNA using superscript II Reverse transcriptase (Invitrogen) General Infirmary, United Kingdom, between 1981 and 2000 were studied. andoligo-dT(T18) primers (Research Biolabs). Semiquantitative reverse One hundred ninety-one (62.4%) were male and 115 (37.6%) were female, with transcription-PCR (RT-PCR) was performedusing RAB23 and h-ACTIN a median age of 72.1 y, ranging from ages 36 to 96 y. Additional specific primers (Research Biolabs): RAB23_fwd, 5¶-GGTACCCTCAAT- clinicopathologic variables, including tumor-node-metastasis classification GGTGGAGA; RAB23_rev, 5¶-ATGACAGCTGGATGGGTTTC; ACTB_fwd, 5¶- (22), tumor differentiation grade according to WHO classification (23), CGGGAAATCGTGCGTGACATTAAG; ACTB_rev, 5¶-TGATCTCCTTCTG- tumor stage, andhistologic subtype according to Lauren classification (10) CATCCTGTCGG. are provided in the Supplementary Data. This study was performed with Western blotting. Cell pellets were lysed in modified radioimmunopre- the approval of the Leeds (West) Research Ethics Committee (LREC No cipitation assay lysis buffer with a cocktail of protease inhibitors (Roche). CA01/122). For each cell line, 30 Ag of protein were separatedon a 15% SDS poly- TMAs and immunohistochemistry. Cell lines were fixed and embedded acrylamide gel and electroblotted onto a polyvinylidene difluoride into paraffin, andusedto construct a cell line array using a Beecher membrane. The blots were probedwith anti-RAB23 (1:1,000) or anti– h- microarrayer (24). TMAs were constructedfrom routine formalin-fixed, ACTIN (1:1,000; Sigma) antibodies, incubated with horseradish peroxidase– paraffin-embedded GC tissue and matched normal mucosa (if available; conjugated secondary antibodies, and visualized using a chemiluminescence ref. 25). Three cores were taken from tumors with high tumor cell density, detection reagent (Bio-Rad). andsix cores from tumors with low tumor cell density(<50% tumor cells

6 http://biosun1.harvard.edu/complab/dchip/ 7 http://www.genedata.com 8 http://genome-www.stanford.edu/gastric_cancer/index.shtml

Cancer Res 2008; 68: (12). June 15, 2008 4624 www.aacrjournals.org

Downloaded from cancerres.aacrjournals.org on October 3, 2021. © 2008 American Association for Cancer Research. RAB23 Expression in Diffuse-Type Gastric Cancer per area). Each TMA also containedseveral non-gastric tissue samples to using high-resolution aCGH on two different platforms: a BAC- assess sensitivity andspecificity of the immunohistochemical staining. microarray platform comprising 32,000 BACs providing tiling RAB23 primary antibodies were diluted 1:1,000 in antibody diluent (Zymed resolution coverage of the whole human genome (15), andan Laboratories), andbiotinylatedanti-rabbit (ready-madeDAKO ChemMate oligonucleotide microarray platform containing probes represent- Bottle A; DAKO Cytomation) was usedas a secondary antibody.Further ing 100,000 single-nucleotide polymorphism (SNP)s. We confirmed immunohistochemical details are provided in the Supplementary Data. The intensity of the RAB23 immunohistochemical reaction was scoredas several previously described amplicons in this cell line panel, c-Myc negative, weak, moderate, or strong. To avoid false-negative cases, a tumor such as amplification at chromosome 8q24 in SNU-16 ERBB2 was only categorizedas ‘‘negative’’ if positive staining was present in non- cells, amplification at chromosome 17q21 in NCI-N87 cells, malignant cells of the same core or any other core of the same case. In and c-Met amplification at chromosome 7q31 in SNU-5 cells preliminary work, we foundthat RAB23 was expressedin a cell membrane– (Supplementary Fig. S1). We also discovered a novel high-level focal associatedpattern in all normal gastric epithelial cells but in a amplification at chromosomal region 11q22 in YCC3 andYCC7 predominantly cytoplasmic pattern in tumor cells (see Results and the cells (Supplementary Fig. S1), containing the genes BIRC2 and Supporting Data). In the absence of normal gastric epithelial cells, positively YAP1, recently identified as coamplified genes with cooperative stainedlymphocytes or smooth muscle cells were usedas internal controls. effects in liver cancer (26). Taken collectively, these results support Cores without positive internal controls were excluded from the statistical the validity and robustness of the aCGH data. analyses, resulting in the overall number of informative cases being 306. Statistical analyses. Statistical analyses were done using SPSS 12.0.1 for We identified an amplified region at Chr 6p11p12 at genomic Windows (SPSS, Inc.). For the analyses of RAB23 expression, comparisons position 56 Mb (genome buildNational Center for Biotechnology for the different staining intensity categories were performed using either Information 36.1) in Hs746T cells (Fig. 1A). The enhancedresolution the Mann-Whitney test (for two groups, negative/weak versus moderate/ of our aCGH platform allowedus to narrow downthe amplicon strong) or the Kruskal-Wallis test (for more than two groups). region to f3 Mb, comparedwith the 5.8 Mb obtainedby conventional CGH (data not shown). The presence of this 6p11p12 amplification was confirmedby 100K SNP arrays (Fig. 1 B Results and C, representing chromosomal andclose-in views, respectively). High resolution aCGH identifies a focal chromosomal Excellent concordance was seen between the BAC and SNP arrays amplification at 6p11p12 in GC. To identify regions of chromo- (Fig. 1D), indicating that the same genomic aberration can be somal amplification in GC, we interrogatedseven GC cell lines detected by using different high-resolution array technologies.

Figure 1. Identification of the chromosome 6p amplicon. A, aCGH profiles of chromosome 6 in GC cell lines. Y-axis, log2-transformed LOWESS smoothened values averaged over a dye-swap replicate. X-axis, the physical genomic location along chromosome 6. Black arrow, the region of interest. B, SNP arrayprofiles of chromosome 6. Top red rows, inferred copynumber for samples (median smoothing, window width is nine). Gray box on the bottom, the value range from 0 to 4 copies; red line, a normal copynumber of 2. Blue curve in the gray box, the copynumber of the Hs746T cell line. C, close-up views of inferred copynumber for the amplified locus bySNP array(median smoothing, window width is nine). D, zoom-in diagrams for aCGH and SNP arrayprofiles. Smoothened and unsmoothened data from replicates (colored dots) are shown for the cell line Hs746T. Red bars, peak region of amplification. www.aacrjournals.org 4625 Cancer Res 2008; 68: (12). June 15, 2008

RAB23, a candidate 6p11p12 amplicon gene, is amplified and overexpressed in Hs746T cells. By integrating the BAC and SNP array results, we localizedthe peak of the 6p11p12 amplification to 57.1 and57.4 Mb (Fig. 1 D, red bar). Two candidate genes whose genomic coordinates were entirely encompassed within the amplification peak were identified, BCL2-associated athanogene 2 (BAG2; Chr6: 57,145,293-57,157,694), and RAB23 (Chr6: 57,161,541-57,194,216), a small GTPase belonging to the Ras superfamily (Fig. 2A). Although BAG2 and RAB23 lie in extremely close proximity to one another, being separatedonly by f4 kb, the two genes are transcribed in opposite directions, indicating that they possess distinct promoters. Besides BAG2 and RAB23,two other genes were partially coveredby the amplification peak— ZNF451, a zinc finger protein, and PRIM2A, a large primase subunit involvedin DNA replication. DST, a member of the plakin protein family foundat adhesionjunctions, was locatedclose to this region but fell outside the predicted amplification peak. To investigate if genes within the 6p11p12 amplicon might exhibit ‘‘copy number–driven expression,’’ we analyzed gene expression values for these candidate genes from a gene expression profile database of GC cell lines (18). RAB23 was overexpressedin Hs746T by >4.5-foldrelative to the median expression of all 7 GC cell lines (Supplementary Fig. S2). Other 6p11p12 genes involvedin the predictedregion of amplification were also overexpressedbut to a lesser extent than RAB23. These include BAG2, followedby PRIM2A, ZNF451, and DST. Notably, the expression levels of other 6p11p12 genes outside the amplified region (e.g., HCRTR2, BMP5, and KHDRBS2) were not significantly different between Hs746T and the other cell lines (Supplementary Fig. S2), supporting the notion that the elevatedexpression levels of genes within this region in Hs746T cells is likely due to DNA amplification. The relatively high level of RAB23 overexpression motivatedus to further investigate the potential role of this gene in GC. Validation of RAB23 overexpression—gene copy number, mRNA, and protein expression. To confirm that the 6p11p12 region was indeed amplified in Hs746T cells, we conducted FISH experiments using chr6p11p12-specific probes (RP11-641O11) and a reference centromere 6 probe (D6Z1; Fig. 2B). In SNU-1 cells that did not show a 6p11p12 amplification by aCGH, we consistently observeda diploidratioof 6p11p12 to centromere 6 signals (Fig. 2B). In contrast, in Hs746T cells, the modal number of 6p11p12 signals was 8, whereas the modal centromere 6 number was 5, indicating that the 6p11p12 region is indeed amplified relative to centromere 6. In Hs746T cells, the 6p11p12 signals Figure 2. RAB23 is amplified and highlyexpressed in Hs746T Cells. A, UCSC localizedto multiple scatteredareas across the cell nucleus in Genome Browser on human (genome build NCBI 36.1). Red bar, peak region of amplification. Genes falling entirelywithin the peak ( RAB23 and BAG2) contrast to being colocalizedwith the centromere 6 probe as in are highlighted in red. B, validation of the chromosome 6p11p12 amplicon by SNU-1 cells (Fig. 2B), suggesting that this region has translocatedto FISH (Â100). Hs746T cells (100; left) and SNU-1 cells (right) were analyzed. multiple regions in different chromosomes. All SNU-1 cells counted registered a pair of signals. C, verification of RAB23 RAB23 gene amplification bygenomic qPCR. The gene copynumber in Hs746T cells To confirm amplification at the single gene resolution, we was compared with other cell lines. All RAB23 values were normalized against quantified RAB23 gene copy numbers in these lines by genomic an ACTB control. D, confirmation of high RAB23 gene expression in Hs746T qPCR. Hs746T cells exhibiteda four to five times increasedgene cells byRT-PCR ( top) and RAB23 protein expression byWestern blotting (WB; bottom). Note that besides HS746T cells, AGS and SNU-1 cells also copy number of RAB23 relative to a h-actin (ACTB) control gene express RAB23. (Fig. 2C). Semiquantitative RT-PCR showedthat RAB23 (Fig. 2D) was significantly overexpressedin Hs746T cells comparedwith other cell lines confirming the genome-wide Affymetrix expression RAB23 protein was also expressedin SNU-1 andAGS cells albeit at array results. Furthermore, Western immunoblotting using a lower level than in Hs746T cells; SNU-16 cells hadno detectable RAB23 antibodies showed that RAB23 was overexpressed in RAB23 expression; andNCI-N87, KATO III, andSNU-5 cells had Hs746T cells comparedwith other cell lines (Fig. 2 D). Interestingly, very weak RAB23 expression. These findings suggest that in the immunoblotting results also revealeda considerable level of addition to gene amplification, RAB23 expression in some cell lines heterogeneity in RAB23 protein expression in the other lines. may be modulated by other mechanisms.

Cancer Res 2008; 68: (12). June 15, 2008 4626 www.aacrjournals.org

A potential cell invasion role for RAB23 in amplified and off-target effect. To test if the trends seen in the Matrigel nonamplified cell lines. To investigate the potential function of experiments could also be observed in an independent assay, we RAB23, we performed siRNA-mediated knockdown experiments. performedcell migration ‘‘woundhealing’’ assays (see Materials Using a pooledset of siRNAs targeting RAB23, we successfully andMethods). Treatment of Hs746T cells with either RAB23 siRNA- silenced RAB23 at the mRNA andprotein level in RAB23-amplified 5orRAB23 siRNA-6 both resultedin a significant inhibition of cell Hs746T cells (Supplementary Fig. S3; Fig. 3A). We also included in migration (Supplementary Fig. S4; Fig. 3C; P = 0.0097 and0.003 for these experiments siRNAs targeting BAG2 (Supplementary Fig. S3). RAB23 siRNA-5 andsiRNA-6, respectively). These results support a We foundthat there was no significant difference in cell functional role for RAB23 as an invasion mediator gene in GC. proliferation capacity between control cells, RAB23-, BAG2-, and We then investigatedif the invasion mediatoreffects of RAB23 RAB23/BAG2-silencedcells (Supplementary Fig. S4). However, in a might also be observedin other lines. No effects of RAB23 siRNA Matrigel invasion assay, both RAB23-silencedand RAB23/BAG2– were observedin non- RAB23–expressing cell lines such as YCC3 double silenced Hs746T cells exhibited a decreased level of cell andYCC10 (Supplementary Fig. S3; datanot shown). In AGS cells, invasion comparedwith control cells (Fig. 3 A and B; P = 0.01 and which express RAB23 but do not carry RAB23 amplifications 0.005 for RAB23 siRNA and RAB23/BAG2 siRNA, respectively), (Supplementary Fig. S3; Fig. 2D), RAB23 knockdown resulted in a suggesting a potential role for RAB23 in cell motility andinvasion. significant reduction of invasiveness similar to HS746T cells In contrast, BAG2 silencing alone did not significantly affect (Supplementary Fig. S4; Fig. 3D; P = 0.028). To extendthis finding, cellular invasion (Fig. 3B). we askedif RAB23 overexpression might prove sufficient to induce To rule out potential off-target effects of the RAB23 pooled invasiveness in AGS cells. Using a RAB23 cDNA expression siRNA, we repeatedthe Matrigel invasion assays using two construct, we foundthat RAB23 overexpression resultedin a independent siRNAs (siRNA-5 and siRNA-6) targeting distinct significant enhancement of invasion (Fig. 3D; P = 0.00072). Taken parts of the RAB23 gene (Supplementary Fig. S3). The efficacy of collectively, these results suggest that RAB23 expression may the single siRNA treatments were confirmedby Western blotting influence the invasive capacity of a variety of different GC cells. (Supplementary Fig. S3). Similar to the pooledsiRNA experiments, RAB23 expression is significantly associated with diffuse- treating Hs746T cells with either RAB23 siRNA-5 or RAB23 siRNA-6 type GC in multiple patient cohorts. To investigate RAB23 in also resultedin a significant reductionof invasion (Supplementary primary cancers, we performedFISH on primary gastric tumors. Fig. S4; P = 0.0053 and0.0011). These results indicatethat the We detected RAB23 amplification in 2 of 10 primary GCs (Fig. 4A). effects of the RAB23 siRNA treatment are unlikely to be due to an Of the remaining eight, five tumors exhibiteda normal number of

Figure 3. Functional studies of RAB23. A, confirmation of gene silencing in Hs746T cells byboth RT-PCR (mRNA) and Western blotting (protein). Right, representative images of the center field of invasion chambers for control-transfected and RAB23-silenced Hs746T cells. RAB23 siRNA resulted in significantlyreduced numbers of invading cells. Red arrows, representative cells invading through the transwell, which should be distinguished from the 8-Am pores (small round circles). Bars, 40 Am. B, invasion assays in Hs746T cells after siRNA treatment. RAB23 siRNA and combined RAB23/BAG2 siRNA significantlyinhibited the invasive potential of Hs746 cells (red P values), whereas BAG2 knockdown had onlymarginal effects. C, migration assays in Hs746T cells after siRNA treatment. Hs746T cells treated with two distinct RAB23 siRNAs (siRNA-5 and siRNA-6) exhibited a significant decrease in migration capacity compared with control siRNA–treated cells after 48 h (red star). Significant differences in migration capacitywere alreadyobserved at 24 h ( P <0.05for both siRNA-5 and siRNA-6). D, invasion assays in RAB23-silenced and RAB23-overexpressing AGS cells. Left, treatment of AGS cells with RAB23 siRNA significantlyinhibited the invasive potential of AGS cells (P = 0.028). Right, AGS cells overexpressing a RAB23 cDNA (AGS-RAB23) exhibited significantly enhanced invasion compared with cells expressing an AGS-pCl-Neo control vector (P = 0.00072). The Western blot insert confirms ectopic RAB23 overexpression in AGS cells. All results were averaged over three independent experiments.

www.aacrjournals.org 4627 Cancer Res 2008; 68: (12). June 15, 2008

Figure 4. RAB23 gene, transcript, and protein expression in primarytumors. A, FISH analysis of primary tumors. Gastric tumor imprints were probed with a RAB23 6p11 probe, RP11-641O11 (green), and a D6Z1 centromeric 6 probe (red). Left, gastric tumor cells exhibiting 6p11 amplification (multiple green signals) with a normal set of centromere 6 signals (two copies). Right, normal gastric epithelial cells from the same patient, exhibiting two copies of both RAB23 and centromere 6. 4¶,6-diamidino-2-phenylindole was used to counterstain the nucleus. B, genomic qPCR of RAB23 gene in 70 GC samples [Singapore (SG) cohort]. Red columns, tumors exhibiting a RAB23/ACTB ratio of >1.25; dotted line, a threshold RAB23/ACTB cutoff of >1.5. Three to five tumors meet this latter threshold. C, gene expression profiles of RAB23 gene in 90 GC samples [Hong Kong (HK) cohort; ref. 27]. Dotted line, RAB23 gene expression that is >2-fold above than the mean RAB23 expression level in normal gastric tissues. Red columns, tumors that pass this criterion. D, RAB23 immunohistochemistryin United Kingdom cohort. Hs746T, strong diffuse cytoplasmic immunoreactivity; KATO III, weak cytoplasmic immunoreactivity prominently in intracytoplasmic vesicle–like structure and cell membrane; dGC, strong diffuse cytoplasmic immunoreactivity of RAB23 in dGC; iGC, weak to no RAB23 immunoreactivityin iGC. Bars, 40 Am.

RAB23 andchromosome 6 signals, two displayedconcomitantly nine normal gastric tissues and15 gastric tumors from Singapore. increasednumbers of RAB23 andchromosome-specific signals Using a cutoff thresholdof >2-foldexpression in cancers suggesting gross aneuploidy, and one tumor exhibited increased comparedwith the median RAB23 expression level in normal chromosome 6 signals comparedwith RAB23. To extendthese gastric tissues, 7 of 15 tumors (46%) overexpressed RAB23 findings to a larger sample series, we used genomic qPCR to (Supplementary Fig. S5). To extendthese findingsfrom the quantify RAB23 gene copy number in 70 primary GC samples from relatively small Singapore data set to a larger series with known Singapore. We first measured RAB23 copy numbers relative to an histopathologic andclinical variables, we investigateda publicly ACTB genomic control in 4 diploid reference genomic DNAs and available data set containing the gene expression profiles of 90 founda qPCR mean of 1.1-foldwith a SD of 0.1—thus, measure- GCs from Hong Kong (27). Using the same thresholdcutoff as the ments above 1.3-foldare likely to represent true deviationsfrom quantitative RT-PCR analysis (>2-foldgreater expression in tumors normal DNA copy number at 95% confidence. Under these comparedwith normal tissues), f35% of GCs in the Hong Kong conditions, we identified 9 tumors (13%) exhibiting potential cohort overexpressed RAB23 at the transcript level (Fig. 4C), a RAB23 expression gains with RAB23/h-actin ratios of >1.25. Of percentage comparable with the Singapore samples. Notably, the these, 3 to 5 tumors likely harbor RAB23 amplifications with observation that more cancers overexpress RAB23 than show RAB23/h-actin ratios of z1.5 (Fig. 4B). Taken collectively, these RAB23 gene amplification is consistent with the results we results show that RAB23 amplifications can indeed occur in obtainedfrom GC cell lines. No significant associations were primary GC. foundbetween RAB23 expression andtumor site or tumor stage To compare RAB23 gene expression between normal gastric (Supplementary Table S1). However, RAB23 expression level was tissues andcancers, we performed RAB23 quantitative RT-PCR on significantly associatedwith histologic tumor type, with dGCs

Cancer Res 2008; 68: (12). June 15, 2008 4628 www.aacrjournals.org

Downloaded from cancerres.aacrjournals.org on October 3, 2021. © 2008 American Association for Cancer Research. RAB23 Expression in Diffuse-Type Gastric Cancer exhibiting significantly higher expression levels of RAB23 com- (Supplementary Fig. S7). Furthermore, it shouldbe remembered paredwith iGCs (Spearson m2, P = 0.01). that other cellular mechanisms besides gene amplification may also To validate the association between RAB23 expression and dGC cause overexpression of these genes. This was confirmedin our at the protein level, we performedimmunohistochemistry in a study, where RAB23 protein expression was also discovered in third independent cohort of >300 GC patients from the United several non-RAB23–amplifiedcell lines (AGS andSNU-1; Fig. 2 D), Kingdom. First, we used cell lines to confirm the specificity and andin primary GCs, a larger fraction of tumors overexpressed RAB23 sensitivity of the RAB23 antibody in formalin-fixed paraffin- at the transcript level comparedwith the fraction with RAB23 gene embedded material—similar to the immunoblotting data, different amplifications (Fig. 4A and B versus C). This apparent discordance cell lines expresseddifferent levels of RAB23, with RAB23 staining between RAB23 gene andprotein is unlikely to be specific to being strongest in Hs746T, andweakest in KatoIII cells (Fig. 4 D). In RAB23—it is well-establishedthat there exists an imperfect these lines, RAB23 was also predominantly located in cytoplasmic correlation between gene andprotein expression (29, 30). This may dots suggestive of being located in vesicles and was sometimes be due to the activity of additional as-yet-unidentified regulatory observedat the cell membrane (Fig. 4 D). Second, RAB23 IHC was mechanisms that differentially modulate transcript versus protein informative in 306 primary gastric adenocarcinomas and normal levels. However, this observation does not detract from our main gastric mucosa. One hundred seventy-seven (58%) GCs showed conclusion that identifying genes under focal amplifications in cell moderate to strong immunoreactivity, and 129 (42%) GCs showed lines, with subsequent validation in primary tumors, may prove a weak or no immunoreactivity for RAB23. Although normal gastric useful general strategy to identify new cancer related genes. epithelial cells showeda membranous staining pattern with very RAB23 is locatedat chromosome position 6p11p12, and little cytoplasmic staining (Supplementary Fig. S6), RAB23 in chromosome 6p amplifications have been previously reportedin tumors was most commonly seen in a cytoplasmic ‘‘vesicular’’ and GC using conventional CGH (2, 8). However, although RAB23 was the sometimes cytoplasmic diffuse pattern similar to the pattern in cell most overexpressed candidate gene in this 6p region, it is possible lines (Fig. 4D). Consistent with mRNA transcript findings in the that other chromosome 6p genes, particularly those flanking RAB23, Hong Kong data set, dGCs in the UK data set showed significantly may also play a role in GC. Indeed, we also investigated the function higher RAB23 immunoreactivity comparedwith iGC ( P = 0.003; of BAG2 as this gene is locatedin close proximity to RAB23.We Supplementary Table S2; Fig. 4D). No association of RAB23 foundthat reducing BAG2 expression by siRNA did not result in a expression was found with gender, depth of tumor invasion (pT), significant reduction of cell proliferation, invasion, or apoptosis lymph node status (pN), tumor stage, and grade of differentiation (Supplementary Fig. S2; Fig. 4; data not shown), suggesting that it (Supplementary Table S1). These results, observedat both the gene may be a bystander gene. However, more work will be required to expression and protein level in two independent patient cohorts assess the potential role of other 6p11p12 genes in GC. (Hong Kong andthe UnitedKingdom),confirm that RAB23 RAB23 is a member of the Rab GTPase family (31). Although our expression is significantly andconsistently associatedwith dGC. functional analysis suggests a role for RAB23 in cancer cell invasion, the exact cellular mechanism of RAB23 is currently unclear. RAB23 may play a role in facilitating vesicular transport, controlling Discussion endocytic progression to lysosomes (32). Interestingly, although In this study, we used integrative genomics to identify novel RAB23 was originally reportedto be predominantly expressedin genes involvedin GC. By interrogating a panel of GC cell lines with the brain (32, 33), RAB23 has also recently been implicatedin two different high-resolution aCGH platforms and gene expression carcinogenesis. High RAB23 expression was reportedin hepatocel- profiling, we identified and subsequently validated RAB23 as an lular carcinoma patients andassociatedwith tumor size (34). More amplifiedandoverexpressedgene in Hs746T cells. Functionally, pertinent to our study, another very recent report identified RAB23 siRNA-mediated knockdown of RAB23 inhibitedthe invasive as an up-regulatedgene in nonmalignant diseasedgastric tissues capacity of both RAB23-amplified(Hs746T) andnonamplified (e.g., atrophic gastritis with intestinal metaplasia) comparedwith (AGS) cells, suggesting that RAB23 may play a role in cancer cell normal gastric mucosa (35). Because atrophic gastritis may increase invasion. We confirmedthe presence of RAB23 amplifications in the risk of GC, these findings raise the possibility that RAB23 up- primary tumors, and in two independent cohorts from Hong Kong regulation may be involvedat an early stage of gastric carcinogensis. andthe UnitedKingdom, RAB23 transcript andRAB23 protein Furthermore, our discovery that RAB23 is both proinvasive and expression was associatedwith tumor morphology (i.e., dGC). significantly associatedwith dGCis particularly intriguing because Our experimental approach was specifically targetedtowardhigh- dGCs are phenotypically more invasive than iGCs. A preliminary level focal genomic amplifications, rather than the more commonly survival analysis of patients expressing either high or low levels of observedlarge genomic aberrations in primary tumors. Compared RAB23 in tumors did not exhibit a statistically significant survival with lower resolution technologies, the boundaries and central peaks difference (data not shown). Nevertheless, our findings provide of these focal amplifications are more robustly detected using high- further support that iGCs anddGCs are likely to represent resolution aCGH platforms due to their higher level of probe molecularly distinct entities, which may prove important for the coverage. Indeed, similar strategies targeting focal high-level development of subtype-specific targeted therapies. amplifications have also been informative in recent studies of liver What are the signaling pathways that RAB23 might regulate in (26) andoral cancer (28). One potential disadvantage of this strategy GC? In the neural system, RAB23 may negatively regulate the sonic is that such focal amplifications are usually rarer comparedwith hedgehog signaling (SHH) pathway (36–38) by controlling the broadamplifications and,thus, might be easily missedif only small subcellular localization of essential SHH components (39, 40). In numbers of cell lines or samples are evaluated. This was observed in this regard, our finding that RAB23 is both amplifiedand our study, where the 6p11p12 region was amplified in only one of the overexpressedin GC may seem counterintuitive, as hyperactivation seven GC cell lines. However, we have subsequently observed of SHH signaling has been shown to be tumorigenic in various amplification of this region in an expanded GC cell line panel cancer types such as small cell lung cancer (41), colorectal www.aacrjournals.org 4629 Cancer Res 2008; 68: (12). June 15, 2008

Downloaded from cancerres.aacrjournals.org on October 3, 2021. © 2008 American Association for Cancer Research. Cancer Research adenocarcinoma (42), medulloblastoma (43), and even GC itself and larger patient cohorts. Furthermore, we plan to address if a (44, 45). However, SHH signaling has also been shown to be molecular classification of GC as at least two separate diseases may important in promoting parietal cell differentiation in the stomach facilitate our ability to identify novel treatment strategies for this (46, 47), and it is thus possible that a disruption of this differ- deadly disease. entiation pathway may be requiredfor GC development.The potential role of RAB23 in influencing SHH signaling in GC deserves Disclosure of Potential Conflicts of Interest to be further examined. Alternatively, increased RAB23 expression No potential conflicts of interest were disclosed. (through gene amplification or other mechanisms) may influence cell migration and invasion in a manner independent of SHH signaling but still relevant to tumorigenesis. Acknowledgments In conclusion, we have successfully shown that an integrative Received10/15/2007; revised4/2/2008; accepted4/16/2008. Grant support: The Singapore Cancer Syndicate (SCS-BS0001) and the Biomedical genomic approach targeting focal high-level amplifications in Research Council (BMRC 05/1/31/19/423; P. Tan), a Ministry of Education AcRF tier 2 cancer cell lines can leadto the identificationof genes important in grant (B.L. Tang), andthe Pathological Society of Great Britain andIrelandandthe GC such as RAB23. Future work will needto focus on identifying NCRI Experimental Cancer Medicine Centre Initiative (H. Grabsch). The costs of publication of this article were defrayed in part by the payment of page additional focal amplifications using even higher resolution aCGH charges. This article must therefore be hereby marked advertisement in accordance platforms (>1 million SNP) on more comprehensive cell line panels with 18 U.S.C. Section 1734 solely to indicate this fact.

References 14. Vidgren V, Varis A, Kokkola A, et al. Concomitant protein andmRNA expression in lung adenocarcinomas. gastrin andERBB2 gene amplifications at 17q12–21 in Mol Cell Proteomics 2002;1:304–13. 1. Grabsch H, Kerr D, Quirke P. Is there a case for the intestinal type of gastric cancer. Genes Chromo- 31. Smith AC, Heo WD, Braun V, et al. A network of Rab routine clinical application of ploidy measurements somes Cancer 1999;24:24–9. GTPases controls phagosome maturation andis mod- in gastrointestinal tumours? Histopathology 2004;45: 15. Ishkanian AS, Malloff CA, Watson SK, et al. A tiling ulatedby Salmonella enterica serovar Typhimurium. 312–34. resolution DNA microarray with complete coverage of J Cell Biol 2007;176:263–8. 2. Sakakura C, Mori T, Sakabe T, et al. Gains, losses, and the human genome. Nat Genet 2004;36:299–303. 32. Guo A, Wang T, Ng EL, et al. Open brain gene product amplifications of genomic materials in primary gastric 16. van der Woude CJ, Kleibeuker JH, Tiebosch AT, et al. RAB23: expression pattern in the adult mouse brain and cancers analyzedby comparative genomic hybridiza- Diffuse andintestinal type gastric carcinomas differ in functional characterization. J Neurosci Res 2006;83: tion. Genes Chromosomes Cancer 1999;24:299–305. their expression of apoptosis relatedproteins. J Clin 1118–27. 3. Kimura Y, Noguchi T, Kawahara K, et al. Genetic Pathol 2003;56:699–702. 33. Olkkonen VM, Peterson JR, Dupree P, et al. Isolation alterations in 102 primary gastric cancers by compar- 17. Keller G, Hofler H, Becker KF. Molecular medicine of a mouse cDNA encoding RAB23, a small novel ative genomic hybridization: gain of 20q and loss of 18q of gastric adenocarcinomas. Expert Rev Mol Med 2005;7: GTPase expressedpredominantlyin the brain. Gene are associatedwith tumor progression. ModPathol 1–13. 1994;138:207–11. 2004;17:1328–37. 18. Aggarwal A, Leong SH, Lee C, et al. Wavelet 34. Liu YJ, Wang Q, Li W, et al. RAB23 is a potential 4. Kang JU, Kang JJ, Kwon KC, et al. Genetic alterations in transformations of tumor expression profiles reveals a biological target for treating hepatocellular carcinoma. primary gastric carcinomas correlatedwith clinicopath- pervasive genome-wide imprinting of aneuploidy on the WorldJ Gastroenterol 2007;13:1010–7. ological variables by array comparative genomic hy- cancer transcriptome. Cancer Res 2005;65:186–94. 35. Kim KR, Oh SY, Park UC, et al. [Gene expression bridization. J Korean Med Sci 2006;21:656–65. 19. Heidemann J, Ogawa H, Dwinell MB, et al. Angio- profiling using oligonucleotide microarray in atrophic 5. Fukuda Y, Kurihara N, Imoto I, et al. CD44 is a genic effects of interleukin 8 (CXCL8) in human gastritis andintestinal metaplasia]. Korean J Gastro- potential target of amplification within the 11p13 intestinal microvascular endothelial cells are mediated enterol 2007;49:209–24. amplicon detected in gastric cancer cell lines. Genes by CXCR2. J Biol Chem 2003;278:8508–15. 36. Evans TM, Simpson F, Parton RG, et al. Character- Chromosomes Cancer 2000;29:315–24. 20. Nam S, Kim D, Cheng JQ, et al. Action of the Src ization of RAB23, a negative regulator of sonic hedgehog 6. Leung SY, Ho C, Tu IP, et al. Comprehensive analysis of family kinase inhibitor, dasatinib (BMS-354825), on signaling. Methods Enzymol 2005;403:759–77. 19q12 amplicon in human gastric cancers. ModPathol human prostate cancer cells. Cancer Res 2005;65:9185–9. 37. Evans TM, Ferguson C, Wainwright BJ, et al. RAB23, a 2006;19:854–63. 21. Albini A andBenelli R. The chemoinvasion assay: a negative regulator of hedgehog signaling, localizes to the 7. Sen S, Zhou H, White RA. A putative serine/threonine methodto assess tumor andendothelial cell invasion plasma membrane andthe endocyticpathway. Traffic kinase encoding gene BTAK on chromosome 20q13 is andits modulation. Nat Protoc 2007;2:504–11. 2003;4:869–84. amplifiedandoverexpressedin human breast cancer 22. Sobin LH, Fleming ID. TNM Classification of 38. Eggenschwiler JT, Espinoza E, Anderson KV. RAB23 is cell lines. Oncogene 1997;14:2195–200. Malignant Tumors, fifth edition ([1997]). Union Inter- an essential negative regulator of the mouse Sonic 8. Nakanishi M, Sakakura C, Fujita Y, et al. Genomic nationale Contre le Cancer andthe American Joint hedgehog signalling pathway. Nature 2001;412:194–8. alterations in primary gastric cancers analyzedby Committee on Cancer. Cancer 1997;80:1803–4. 39. Eggenschwiler JT, Bulgakov OV, Qin J, et al. Mouse comparative genomic hybridization and clinicopatho- 23. Hamilton SR andAaltonen LA, editors. WHO RAB23 regulates hedgehog signaling from smoothened logical factors. Hepatogastroenterology 2000;47:658–62. Classification of Tumours. Pathology andGenetics of to Gli proteins. Dev Biol 2006;290:1–12. 9. Gorringe KL, Boussioutas A, Bowtell DD. Novel regions Tumours of Digestive System. Lyon: IARC press;2002. 40. Wang Y, Ng EL, Tang BL. RAB23: what exactly does it of chromosomal amplification at 6p21, 5p13, and12q14 24. Waterworth A, Hanby A, Speirs V. A novel cell array traffic? Traffic 2006;7:746–50. in gastric cancer identified by array comparative technique for high-throughput, cell-basedanalysis. In 41. Chi S, Huang S, Li C, et al. Activation of the hedgehog genomic hybridization. Genes Chromosomes Cancer Vitro Cell Dev Biol Anim 2005;41:185–7. pathway in a subset of lung cancers. Cancer Lett 2006; 2005;42:247–59. 25. Simon R, Mirlacher M, Sauter G. Tissue microarrays. 244:53–60. 10. Lauren P. The two histological main types of gastric Biotechniques 2004;36:98–105. 42. Bian YH, Huang SH, Yang L, et al. Sonic hedgehog- carcinoma: diffuse and so-called intestinal-type carci- 26. Zender L, Spector MS, Xue W, et al. Identification Gli1 pathway in colorectal adenocarcinomas. World J noma. an attempt at a histo-clinical classification. Acta andvalidation of oncogenes in liver cancer using Gastroenterol 2007;13:1659–65. Pathol Microbiol Scand1965;64:31–49. an integrative oncogenomic approach. Cell 2006;125: 43. Marino S. Medulloblastoma: developmental mecha- 11. Zheng H, Takahashi H, Murai Y, et al. Pathobiological 1253–67. nisms out of control. Trends Mol Med 2005;11:17–22. characteristics of intestinal anddiffuse-type gastric 27. Leung SY, Chen X, Chu KM, et al. Phospholipase A2 44. Katoh Y, Katoh M. Hedgehog signaling pathway and carcinoma in Japan: an immunostaining study on the group IIA expression in gastric adenocarcinoma is gastric cancer. Cancer Biol Ther 2005;4:1050–4. tissue microarray. J Clin Pathol 2007;60:273–7. associatedwith prolongedsurvival andless frequent 45. Ma X, Chen K, Huang S, et al. Frequent activation of 12. Kokkola A, Monni O, Puolakkainen P, et al. 17q12–21 metastasis. Proc Natl AcadSci U S A 2002;99:16203–8. the hedgehog pathway in advanced gastric adenocarci- amplicon, a novel recurrent genetic change in intestinal 28. Snijders AM, Schmidt BL, Fridlyand J, et al. Rare nomas. Carcinogenesis 2005;26:1698–705. type of gastric carcinoma: a comparative genomic amplicons implicate frequent deregulation of cell fate 46. Stepan V, Ramamoorthy S, Nitsche H, et al. hybridization study. Genes Chromosomes Cancer 1997; specification pathways in oral squamous cell carcinoma. Regulation andfunction of the sonic hedgehogsignal 20:38–43. Oncogene 2005;24:4232–42. transduction pathway in isolated gastric parietal cells. 13. Kong G, Oga A, Park CK, et al. DNA sequence copy 29. Gygi SP, Rochon Y, Franza B, et al. Correlation J Biol Chem 2005;280:15700–8. number aberrations associatedwith histological sub- between protein andmRNA abundancein yeast. Mol 47. van den Brink GR, Hardwick JC, Tytgat GN, et al. types andDNA ploidyin gastric carcinoma. Jpn J Cancer Cell Biol 1999;19:1720–30. Sonic hedgehog regulates gastric gland morphogenesis Res 2001;92:740–7. 30. Chen G, Gharib TG, Huang CC, et al. Discordant in man andmouse. Gastroenterology 2001;121:317–28.

Cancer Res 2008; 68: (12). June 15, 2008 4630 www.aacrjournals.org

Downloaded from cancerres.aacrjournals.org on October 3, 2021. © 2008 American Association for Cancer Research. Integrative Genomics Identifies RAB23 as an Invasion Mediator Gene in Diffuse-Type Gastric Cancer

Qingsong Hou, Yong Hui Wu, Heike Grabsch, et al.

Cancer Res 2008;68:4623-4630.

Updated version Access the most recent version of this article at: http://cancerres.aacrjournals.org/content/68/12/4623

Cited articles This article cites 46 articles, 10 of which you can access for free at: http://cancerres.aacrjournals.org/content/68/12/4623.full#ref-list-1

Citing articles This article has been cited by 8 HighWire-hosted articles. Access the articles at: http://cancerres.aacrjournals.org/content/68/12/4623.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://cancerres.aacrjournals.org/content/68/12/4623. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.