Rab25 Is a Tumor Suppressor Gene with Antiangiogenic and Anti-Invasive Activities in Esophageal Squamous Cell Carcinoma

Published OnlineFirst September 18, 2012; DOI: 10.1158/0008-5472.CAN-12-1269

Cancer Tumor and Stem Cell Biology Research

Man Tong1, Kwok Wah Chan1, Jessie Y.J. Bao3, Kai Yau Wong1, Jin-Na Chen2, Pak Shing Kwan1, Kwan Ho Tang1,LiFu2, Yan-Ru Qin4, Si Lok3, Xin-Yuan Guan2, and Stephanie Ma1

Abstract Esophageal squamous cell carcinoma (ESCC), the major histologic subtype of esophageal cancer, is a devastating disease characterized by distinctly high incidences and mortality rates. However, there remains limited understanding of molecular events leading to development and progression of the disease, which are of paramount importance to defining biomarkers for diagnosis, prognosis, and personalized treatment. By high-throughout transcriptome sequence profiling of nontumor and ESCC clinical samples, we identified a subset of significantly differentially expressed genes involved in integrin signaling. The Rab25 gene implicated in endocytic recycling of integrins wastheonlygeneinthisgroupsignificantly downregulated, and its downregulation was confirmed as a frequent event in a second larger cohort of ESCC tumor specimens by quantitative real-time PCR and immunohistochemical analyses. Reduced expression of Rab25 correlated with decreased overall survival and was also documented in ESCC cell lines compared with pooled normal tissues. Demethylation treatment and bisulfite genomic sequencing analyses revealed that downregulation of Rab25 expression in both ESCC cell lines and clinical samples was associated with promoter hypermethylation. Functional studies using lentiviral-based overexpression and suppression systems lent direct support of Rab25 to function as an important tumor suppressor with both anti-invasive and -angiogenic abilities, through a deregulated FAK–Raf–MEK1/2–ERK signaling pathway. Further characterization of Rab25 may provide a prognostic biomarker for ESCC outcome prediction and a novel therapeutic target in ESCC treatment. Cancer Res; 72(22); 1–12. 2012 AACR.

Introduction Linzhou and nearby cities in Henan Province of Northern Esophageal cancer ranks as the sixth leading cause of China, ESCC constitutes more than 90% of all esophageal cancer-related deaths worldwide, with distinctly high inci- cancer cases in the area and has the highest incidences and dences and mortality rates particularly in East Asia, Africa, mortality rates of esophageal cancer reported in the world (1, and North America (1). Esophageal squamous cell carcinoma 2). Despite advances in diagnostic techniques and therapeutic (ESCC) is the most common form of esophageal cancer. The modalities, ESCC remains a devastating malignancy due to late disease is characterized by regional variation in incidences. diagnoses and the aggressive nature of the disease. A better More than 50% of all ESCC cases in the world occur in China. In understanding of the recurrent genetic alterations and under- lying molecular mechanisms involved in ESCC development and progression will facilitate the identification of novel tar- gets, allowing for more sensitive methods of detection, facil- Authors' Affiliations: Departments of 1Pathology and 2Clinical Oncology, 3Genome Research Centre, Li Ka Shing Faculty of Medicine, The University itating earlier diagnosis, and prolonging patient survival. of Hong Kong, Hong Kong; and 4Department of Clinical Oncology, First With the advent of next-generation sequencing technologies fi Af liated Hospital, Zhengzhou University, Zhengzhou, China in recent years, a new sequencing platform, called transcrip- Note: Supplementary data for this article are available at Cancer Research tome sequencing (RNA-Seq), has been applied to delineate Online (http://cancerres.aacrjournals.org/). changes at the transcriptomic level. The development of Accession code: Transcriptome sequencing data have been submitted to ESCCs, like many other cancers, is believed to be driven by Gene Expression Omnibus (GEO, http://www.ncbi.nlm.nih.gov/geo/) under the accumulation of genetic alterations, causing the transfor- the accession number GSE32424. mation of normal cells to malignant cells. Thus, studying Corresponding Author: Stephanie Ma, Department of Pathology, Fac- recurrent changes at the levels of functional transcripts in ulty of Medicine, The University of Hong Kong, Room 53, 10/F, Labo- ratory Block, Faculty of Medicine Building, 21 Sassoon Road, Pok Fu malignant cells compared with nontumor cells may aid in the Lam, Hong Kong. Phone: 852-2819-9785; Fax: 852-2218-5244; E-mail: identification of deregulated molecular events and pathways [email protected] involved in driving ESCCs. In the present study, we conducted doi: 10.1158/0008-5472.CAN-12-1269 RNA-Seq analysis on 12 patient-derived nontumor and ESCC 2012 American Association for Cancer Research. clinical samples and identified a number of commonly and

www.aacrjournals.org OF1

Tong et al.

significantly differentially expressed genes. Pathway enrich- plementary Table S1 provides a summary of the clinicopath- ment analysis (DAVID databases) found the deregulated genes ologic parameters [i.e., patient age, gender, tumor–node– to be commonly associated with a number of cancer-related metastasis (TNM) grade] of each of the patients collected for pathways. Of these, 2 of the most significantly enriched path- this purpose. Immunohistochemistry for Rab25 was conducted ways are related to integrin signaling, which is commonly on a tissue microarray (TMA) consisting of 270 pairs of known to influence important cellular processes critical to formalin-fixed, paraffin-embedded ESCC tumor and nontumor tumor development and progression, including cell prolifera- specimens (15). tion, cell survival, angiogenesis, cell motility, and invasiveness (3–6). Among the differentially expressed genes involved in RNA-Seq and differential expression analysis integrin signaling, Ras-related protein Rab25 was found to be RNA-Seq was conducted on the Illumina Cluster Station and the only significantly downregulated gene in ESCC compared GAIIx using the Standard Cluster Generation Kit v4 and the 36- with nontumor tissue and was thus chosen for further studies. Cycle Sequencing Kit v3. cDNA libraries from 7 ESCC samples Rab25 belongs to the Rab family of small GTPases and plays a (1T, 2T, 3T, 6T, 7T, 8T, 9T) and 5 nontumor samples (4N, 5N, 6N, critical role in the maintenance of normal epithelial lining (7– 8N, 9N) were sequenced with 38-base single reads. Sequencing 9). Past studies have shown Rab25 to play very contrasting roles reads were filtered for polymers, primer adaptors, and ribo- in cancer, depending on the tissue in which it is expressed. It somal RNAs and then mapped against the human genome has previously been implicated in the progression of ovarian assembly (NCBI Build 37.1) using CLC Genomic Workbench. and breast cancer (10–12), whereas in contrast, more recent The expression abundance for each gene was measured by studies have identified a tumor-suppressive role of Rab25 in RPKM (number of exon reads mapped per kilobase per million colon cancer and triple-negative breast cancer (13, 14). To date, mapped reads; ref. 16). The differential expression between the role of Rab25 in ESCCs has not been explored. In the tumor and nontumor samples was evaluated using the t test present study, frequent Rab25 downregulation was identified and Baggerley test (17) by treating the same type of individuals in ESCC clinical specimens compared with its nontumor as one group. The genes with a consistent type of regulation for counterparts, whereas reduced Rab25 expression significantly both the t test and Baggerley test were kept, and the genes with correlated with worst overall survival. Absent or significantly a Bonferroni corrected P less than 0.001 were considered to be weaker expression of Rab25 was also likewise detected in a significant. panel of ESCC cell lines as compared with a pooled normal tissue control. Downregulation of Rab25 expression in both ESCC cell lines and culture conditions ESCC cell lines and clinical samples was found to be signifi- ESCC cell lines EC18 and EC109 were kindly provided by cantly associated with promoter hypermethylation, as evi- Professor George Tsao (Department of Anatomy, The Univer- denced by our results obtained from 5-aza-20-deoxycitidine sity of Hong Kong). ESCC cell lines HKESC1 and KYSE520 were (5-aza-dC) demethylation treatment and bisulfite genomic provided by Professor Gopesh Srivastava (Department of sequencing (BGS). Finally, functional studies found Rab25 to Pathology, The University of Hong Kong). The other 5 ESCC possess both anti-invasive and antiangiogenic abilities through cell lines, KYSE30, KYSE140, KYSE180, KYSE410, and KYSE510, a dysregulated mitogen-activated protein kinase (MAPK)/ were obtained from DSMZ, the German Resource Centre for extracellular signal–regulated kinase (ERK) signaling pathway. Biological Material (18). KYSE30 and KYSE180 cell lines were Taken together, our results suggest Rab25 to function as a cultured in Dulbeccos' Modified Eagles' Media. All other ESCC novel tumor suppressor in ESCCs by repressing invasion, cell lines were maintained in RPMI. Both media were supple- angiogenesis, and tumorigenicity. mented with 10% FBS and 1% penicillin/streptomycin. All cell lines used in this study were regularly authenticated by mor- Materials and Methods phologic observation and tested for absence of Mycoplasma Collection of esophageal tissue samples contamination (MycoAlert, Lonza Rockland). All clinical specimens used for RNA-Seq and Rab25 expression studies by quantitative real-time PCR (qPCR), Western RNA extraction, cDNA synthesis, and qPCR blotting, and immunohistochemistry were collected from Total RNA was isolated using TRIzol reagent (Invitrogen). patients with ESCCs who underwent surgical resection of First-strand cDNA was synthesized from 1 mg of total RNA tumor tissues at Linzhou Cancer Hospital (Henan, China). using the Advantage RT-for-PCR kit (Clontech Laboratories) When available, paired adjacent nontumor tissues from the and was then used for qPCR analysis. qPCR was carried out proximal resection margins (>5 cm away from the ESCC using SYBR Green PCR master mix (Applied Biosystems) on an sample) were also collected. The patients had received no ABI Prism 7900HT System (Applied Biosystems). b-Actin was previous local or systemic treatment before operation. All amplified as an internal control. Supplementary Table S2 clinical samples used in this study were approved by the provides a list of the primer sequence used to amplify Rab25 committee for ethical review of research involving human in the qPCR assay. subjects at Zhengzhou University (Zhengzhou, China) and The University of Hong Kong (Pok Fu Lam, Hong Kong). For RNA- DNA extraction, bisulfite modification, and promoter Seq, 12 fresh-frozen clinical specimens (including 3 paired methylation analysis tumor and nontumor, 4 unpaired tumor, and 2 unpaired Genomic DNA was extracted from normal and tumor nontumor samples) were randomly selected for analysis. Sup- esophageal tissues and cell lines by phenol–chloroform

OF2 Cancer Res; 72(22) November 15, 2012 Cancer Research

Rab25 in Esophageal Cancer

method followed by bisulfite modification using the EpiTECT sented genes, 1,598 genes were upregulated and 132 were Bisulfite Kit (Qiagen). Methylation-specific PCR (MSP) and downregulated with a Bonferroni-corrected P less than 0.001 BGS were conducted as previously described (19) using pri- (red dots in Fig. 1B). The global profile of the differential mers listed in Supplementary Table S2. expression profiling for the 1,730 deregulated genes is shown as a heatmap (Fig. 1C), where the genes are listed in descending Immunohistochemistry order according to the Baggerley test analysis. Subsequent Paraffin sections were deparaffinized in xylene and rehy- pathway enrichment analysis by DAVID found the 1,730 dif- drated in graded alcohols and distilled water. Slides were ferentially expressed genes to be commonly associated with a heated for antigen retrieval in 10 mmol/L citrate (pH 6.0). number of cancer-related pathways. Of these, 2 of the most Sections were incubated with polyclonal rabbit anti-human significantly enriched pathways are related to integrin signal- Rab25 (1:400, gift from Dr. Kwai Wa Cheng, University of Texas ing (Fig. 1D). Supplementary Table S4 provides a full list of the MD Anderson Cancer Center, Houston, TX), rat anti-human significantly differentially expressed genes involved in integrin CD34 (Biogenex), mouse anti-human PCNA (Santa Cruz Bio- signaling pathway (P < 0.001). Among these, Ras-related pro- technology), mouse anti-human cytokeratin 5/6 (Millipore), or tein Rab25 was found to be the only significantly downregu- rabbit anti-human phospho-ERK1/2 (Cell Signaling) antibo- lated gene (P < 1E-20) in ESCC compared with nontumor tissue dies overnight at 4C. EnVision Plus System-HRP (DAB; Dako) (Fig. 1E and Supplementary Fig. S1) and was thus chosen for was used according to manufacturer's instruction. Staining further studies. RPKM values obtained from RNA-Seq were was revealed by counterstaining with hematoxylin. Stained reproducible upon qPCR analysis, showing higher Rab25 slides were scanned by Aperio Scanscope CS system (Vista). expression in normal samples versus tumor samples (Fig. Evaluation of immunohistochemical staining for Rab25 was 1F). The result indicates a high level of concordance of the conducted by pathologist K.W. Chan (Department of Pathol- differential expression measurements between both platforms. ogy, The University of Hong Kong) who had no prior knowledge of patient data. Staining intensity was divided into 3 scores: no Rab25 as a novel tumor suppressor gene in human ESCCs (0% staining), low (<50% staining), or high (>50% staining) To determine whether downregulation of Rab25 was a Rab25 expression. Both relative intensity and number of cells common event in ESCCs, we extended our qPCR analysis to stained were taken into consideration during the scoring an additional 43 paired nontumor/primary ESCC samples. process. The specificity of polyclonal rabbit anti-human Rab25 Rab25 was significantly downregulated in tumor tissues when antibody has been verified and reported (10). Staining inten- compared with adjacent nontumor tissues (paired t test, P < sities of proliferating cell nuclear antigen (PCNA) and phos- 0.0001; Fig. 2A). We also examined the expression of Rab25 at pho-ERK1/2 were quantified using Aperio Spectrum positive the protein level by Western blot and immunohistochemical pixel count algorithm. Microvessel density was assessed as analyses. Of the 6 esophageal tissue samples examined by previously described (20). Western blot analysis, Rab25 expression was consistently lower in the 3 ESCC samples than in the 3 nontumor esophageal Results samples (Fig. 2B). To investigate the clinical significance of Differential gene expression profiling by RNA-Seq Rab25 expression in ESCCs, a TMA comprising 270 paired We sequenced 12 RNA libraries from 7 ESCC tumor and 5 nontumor/ESCC samples, with more detailed patient clinical nontumor samples, generating a total of 91 million 38-bp single data, was used for an IHC study. A strong Rab25 staining was reads and an average of 7.6 million reads per sample (Supple- observed in the nontumor samples (Fig. 2C). More than half of mentary Table S3). A total of 72% of the reads aligned with the all informative nontumor samples displayed a high Rab25 human genome, of which 57% aligned with unique locations in expression level (89 of 171, 52.0%), whereas only a quarter of the genome. Across the 12 samples, a total of 14,351 distant the informative tumor samples showed high Rab25 expression genes (42% of the total number of genes in the reference (42 of 164, 25.6%; Fig. 2D; c2 test, P < 0.001). In a Kaplan–Meier database) were expressed with at least 10 mappable exon survival analysis comparing patients with different Rab25 reads. For each gene, the expression value was measured on expression levels, higher Rab25 expression was significantly the basis of RPKM (16). The Euclidean distance between associated with a longer survival time (log-rank test, P ¼ 0.009). samples was calculated, and a hierarchical clustering tree Among 3 groups of patients with different Rab25 expression based on this distance showed a 2-branch (tumor vs. non- levels, increased survival was observed with enhanced Rab25 tumor) partition (Fig. 1A). The 5 nontumor (N) samples and 7 expression (Fig. 2E). Patients with negative Rab25 expression tumor (T) samples clustered into 2 distinct groups, indicating had the worst prognosis, with a mean overall survival of 27.7 that the gene expression between tumor and nontumor sam- months, whereas patients with low Rab25 expression displayed ples differs significantly. Among the 7 tumor samples, samples a relatively improved survival of 34.3 months. Patients with 1T, 6T, 7T, 8T, and 9T were more closely clustered than when high Rab25 expression had the longest mean survival time of compared with 2T and 3T (Fig. 1A). This observation further 45.2 months. suggests that RNA-Seq is a powerful technique in delineating changes at the transcriptomic level. To identify differentially The Rab25 promoter region is frequently expressed genes of functional importance, the t test and hypermethylated in ESCCs Baggerley test were conducted to address the significance of In addition to expression analysis in clinical samples, Rab25 the deregulated genes in tumor samples. Of the 14,351 repre- expression was also examined at both genomic and proteomic

www.aacrjournals.org Cancer Res; 72(22) November 15, 2012 OF3

Tong et al.

AB E 20 Rab25 8T 6T 9T 20 2T 1T RPKM: 249.23 7T 9N 3T 15 0 20 RPKM: 18.27 9T 10 0 ( P value)

10 20 RPKM: 251.11 6N 5 –Log 0 4N 20 9N RPKM: 42.31 6N 0 6T 8N 0 5N −10 −5051015 20 Log (weighted fold-change) 2 RPKM: 266.04 8N 0 CD 20 RPKM: 98.37 8T Nontumor Tumor Gene expression 0 Cell cycle, mitotic 1,000 2,000 3,000 4,000 5,000 6,000 7,000 8,000 9,000

8 N 9 N 5 N 4 N 6 N 6 T 9 T 1 T 7 T 8 T 2 T 3 T Signaling by Wnt Integrin cell–surface... F 12 10 ECM–receptor interaction 8 Integrin signaling pathway 6 Signaling of HGF receptor 4 2 0 50 100 150 Gene count 0 Relative Rab 25 expression Relative 4N5N6N8N9N 1T 2T 3T 6T 7T 8T 9T Enriched pathway P value Nontumor Tumor

Gene expression 2.64×10−21 NT RPKM T RPKM 400 Cell cycle, mitotic 2.92×10−9 4N 302.14 1T 29.18 300 Signaling by Wnt 1.37×10−4 5N 302.42 2T 49.85 Integrin cell–surface 0.0128 200 interations 6N 251.11 3T 47.87 8N 266.04 6T 42.31 ECM–receptor interaction 0.0057 100 r = 0.912

9N 249.23 7T 67.13 (RNA-Seq) RPKM value Integrin signaling pathway 0.0167 P < 0.0001 8T 98.37 0 0 5 10 15 Signaling of HGF receptor 0.0439 –6.0 0 6.0 9T 18.27 Relative Rab 25 expression (qPCR)

Figure 1. RNA-Seq expression profiling data. A, cladogram clustering diagram based on the gene expression values for 12 ESCC and nontumor samples. B, volcano plot of the P value as a function of weighted fold-change for the 5 nontumor versus 7 tumor samples. Blue dots represent genes not significantly differentially expressed (Bonferroni-corrected P > 0.001) and red dots represent genes significantly differentially expressed (Bonferroni-corrected P 0.001). C, heatmap of expression profiles for the 1,730 genes that showed significant expression changes (1,598 upregulated and 132 downregulated). D, pathway enrichment analysis of significantly differentially expressed genes. E, expression tiling of Rab25 in 3 pairs of matched ESCCs and corresponding nontumor clinical samples. F, qPCR analysis of the relative expression of Rab25 in the original 12 ESCC and nontumor clinical samples used for RNA-Seq. Table shows the RPKM expression values of the 12 samples analyzed by RNA-Seq. Validation study to determine the correlation between qPCR and RNA-Seq data.

levels in 8 esophageal cell lines, by qPCR and Western blot the effects of DNA demethylation and histone acetylation on analyses. Absent or significantly downregulated Rab25 expres- Rab25 expression. As shown in Fig. 3C, qPCR analysis sion was observed in a panel of ESCC cell lines when showed a dose-dependent restoration of Rab25 expression compared with a pooled normal tissue control (5 cases of after demethylation treatment with 5-aza-dC. Treatment nontumor esophageal clinical tissue samples pooled togeth- with the histone acetylation agent TSA or VPA did not er). In particular, Rab25 was found to be completely absent significantly alter Rab25 expression, and combined treat- in the ESCC cell lines EC109 and KYSE520 (Fig. 3A and B). ment with 10 mmol/L 5-aza-dC and 0.2 mmol/L TSA did not Downregulation of tumor suppressor genes in cancers is further enhance Rab25 expression. These studies support the often associated with hypermethylation and histone deace- idea that DNA methylation, but not histone modification, is tylation. To determine whether Rab25 downregulation was involved in Rab25 inactivation in ESCC. associated with epigenetic regulation in ESCCs, EC109 and To substantiate the role of aberrant promoter hypermethy- KYSE520 cells that lacked Rab25 expression were treated lation in Rab25 silencing, we conducted MSP or BGS to with varying concentrations of the DNA methyltransferase investigate the methylation status of the Rab25 promoter inhibitor (5-aza-dC) and/or the histone acetylation agents region. The 10-kb sequence directly upstream of the Rab25 [trichostatin A (TSA) or valproic acid (VPA)] to investigate gene was analyzed by 2 publicly available databases for

OF4 Cancer Res; 72(22) November 15, 2012 Cancer Research

Rab25 in Esophageal Cancer

10 P < 0.0001 43 n = 43 paired cases

8 t 6 Nontumor

4 Average Δ C Average

0 Adjacent NT ESCC tumor

B N28 T28 N33T33 N40 T21 ESCC tumor Rab25

β-Actin

D E 100% 1.0 Rab25 expression High Rab25 No Rab25 25.6% expression Low Rab25 80% 0.8 High Rab25 52.0% Low Rab25 No Rab25 - censored expression 60% 0.6 High Low Rab25 - censored 41.5% High Rab25 - censored No Rab25 40% expression 0.4 Low 29.3%

20% survival Cumulative 0.2

Percentage of cases (%) Percentage 32.9% 18.7% No 0.0 P = 0.009 0% Nontumor Tumor 0204060 Mo

Figure 2. Downregulation of Rab25 in ESCCs. A, Rab25 expression in matched nontumor (NT) and ESCC tumor samples (n ¼ 43) as detected by qPCR. b-Actin was used as an internal control. Values displayed as average DCt values. Boxes in the box plot contain the values between the 25th and 75th percentiles. The lines across the boxes indicate the median. The whiskers extend to the highest values, excluding outliers and extremes. B, Rab25 protein expression in ESCC tumor (T) and nontumor (N) tissues as detected by Western blotting. C, representative IHC staining of Rab25 in ESCC tumor and paired nontumor squamous epithelium tissue from one patient sample. D, bar chart summary of the distribution of different Rab25 expression levels in nontumor versus tumor for all informative cases on the TMA. E, Kaplan–Meier survival analysis comparing the overall survival time of patients with ESCCs with different Rab25 expression levels. potential CpG islands: CpG Island Searcher (http://www.cpgis- ESCC cell lines could be detected in these predicted CG lands.com) and EMBOSS-CpGPlot (http://www.ebi.ac.uk/ dinucleotide–rich regions (data not shown). But BGS at CpG Tools/emboss/cpgplot). Three common CpG islands were I4, which contains the promoter region of Rab25 (21), predicted by both programs, including the regions at 8,217 showed a high density of methylation in Rab25-absent EC109 to 6,690 (CpG I1), 6,689 to 5,717 (CpG I2), and 1,287 to and KYSE520 cell lines. In contrast, methylation was rarely 453 (CpG I3) 50 upstream of Rab25 (Fig. 4A). In addition, detected in the same CpG sites in the Rab25-expressing because a recent study also found the sequence from 173 to KYSE30 cell line (Fig. 3D and E). As compared with untreated þ17 (CpG I4) to contain the core promoter region of the EC109, methylation was signiﬁcantly reduced in EC109 cells Rab25 gene (21), we also investigated the methylation status treated with 5-aza-dC (Fig. 3E). In addition, we also inves- in this region (Fig. 3D). We conducted MSP at CpG I1 and tigated the methylation status in a pair of matched ESCCs BGS at CpG I2 and CpG I3; however, no difference in and nontumor clinical sample. Rab25 expression was found methylation status between Rab25 expressing or absent to be signiﬁcantly downregulated in ESCC as compared with

www.aacrjournals.org Cancer Res; 72(22) November 15, 2012 OF5

Tong et al.

A 20 C 12 9.6 10 EC109 15 7.8 8 6 10 Figure 3. Rab25 promoter region is 4 2.4 1.6 2 2 frequently hypermethylated in 1.2 1 1.5 5 2 10.9 1.1 1.1 1 ESCCs. A and B, measurement of 0 genomic and proteomic Rab25 Relative Rab25 expression Relative Relative Rab25 expression Relative 0 1 1 5 expression levels in a panel of 10 50 0.1 100 100 200 400 ESCC cell lines compared with a Double EC18 Control Control Control Normal EC109 pooled normal tissue control by KYSE30 HKESC1 KYSE140KYSE410KYSE510KYSE520 10 qPCR and Western blotting. C, KYSE520 qPCR analysis of Rab25 ESCC 7.3 8 expression after demethylation B ESCC 6 5 and/or histone acetylation treatment with 5-aza-dC (blue 4 2.3 bars), TSA (pink bars), and/or VPA 1.3 1 1.7 (green bars) in ESCC cell lines Normal HKESC1EC18 EC109 KYSE30 KYSE140KYSE410KYSE510KYSE520 2 1.1 0.9 0.9 1 0.7 1 1 0.8 Rab25 without Rab25 expression. Double 0 m Relative Rab25 expression Relative (gray bars), combined 10 mol/L 5-

β-Actin 1 1 5

10 50 aza-dC and 200 nmol/L TSA 0.1 100 100 200 400 treatment. D, a schematic diagram Double Control Control Control showing the distribution of 4 0 5-aza-dC TSA VPA predicted CpG islands 5 - (μmol/L) (nmol/L) (mmol/L) upstream of the Rab25 gene. The CpG islands cover the regions D CpG I1 CpG I2 CpG I3 CpG I4 8,217 to 6,690 (CpG I1), 6,689 to 5,717 (CpG I2), 1,287 to –10,000 bp –8,000 bp –6,000 bp –4,000 bp –2,000 bp +1 bp 453 (CpG I3), and 173 to þ17 (CpG I4). MSP was conducted at MSP BGS BGS CpG I1 and BGS was conducted at MSP BGS CpG I2, CpG I3, and CpG I4. Speciﬁcally, 23 CpG dinucleotides CpG I4 Rab25 are present at CpG I4. BGS primers (4BGS-F and 4BGS-R) were Rab25 core promoter region (CpG I4) +1 bp designed to amplify the sequence in this region. Transcription factor– binding sites for Sp1 and CRE are predicted within this core promoter region. E, mapping of the methylation status of CpG dinucleotides within CpG I4 in a case of matched ESCC and nontumor samples, Rab25- expressing KYSE30 cells, Rab25- E Nontumor negative KYSE520 and EC109 cells, and EC109 cells treated with Tumor 5-aza-dC by BGS. The percentage of methylation at each CpG KYSE30 dinucleotide is displayed in the pie KYSE520 charts.

EC109

EC109 + 5-aza-dC Percentage of methylation: 0% 25% 50% 75% 100%

nontumor in this case as conﬁrmed by qPCR analysis. ESCC notion that DNA promoter hypermethylation is implicated sample was found heavily methylated (75%–100% methyla- in Rab25 inactivation in human ESCCs. tion)in18of21CGdinucleotidesexamined,whereasthe nontumor sample showed signiﬁcantly lower levels of meth- Rab25 suppresses in vitro migration, invasion, and ylation (0%–25%methylation)in17of21CGdinucleotides angiogenesis in ESCCs examined (Fig. 3E, representative shown). Taken together, Frequent epigenetic silencing of Rab25 in ESCC cell lines and these observations provide strong evidence in support of the human ESCC samples prompted us to further investigate the

OF6 Cancer Res; 72(22) November 15, 2012 Cancer Research

Rab25 in Esophageal Cancer

10 A 8.7 B 8 EC109 KYSE30 6 EVRab25 NTC 849 852 4 Rab25 2 1 1 0.12 0.38 β-Actin 0 400 Relative Rab25 expression Relative EVRab25 NTC 849 852 * * 187 EC109 KYSE30 200 102 108 58 23 cells per field C EC109 KYSE30 of migrated No. 0 EV Rab25 NTC 849 852 EV Rab25 NTC 849 852

EC109 KYSE30 Migration 80 * * 41 33 40 28

Invasion 9 cells per field 10 No. of invaded No.

0 EV Rab25 NTC 849 852

EC109 KYSE30 D 40 ** 25 30 ***

14 20 15 10 10 No. of tubes formed No. EVRab25 NTC 849 852 0 0 EC109 KYSE30 EV Rab25 NTC 849 852

EC109 KYSE30

Figure 4. Rab25 from ESCC cell lines suppresses migration, invasion, and angiogenesis in vitro. Stable repression or overexpression of Rab25 in KYSE30 and EC109 cells, respectively, by lentiviral transduction was confirmed by qPCR (A) and Western blotting (B). C, representative migration and invasion assays in Rab25-overexpressed or -repressed clones as compared with their controls. Bar chart summary of the number of migrated or invaded cells per field (, P < 0.05). D, representative images showing morphologic changes and differential tube-forming abilities of HUVECs following treatment with conditioned media from Rab25-overexpressed or -repressed clones as compared with their controls. Bar chart summary showing the number of tubes formed in each assay (, P 0.01; , P 0.001). function of Rab25 in ESCCs. To assess whether Rab25 might versely, suppressing Rab25 expression resulted in an oppos- possess a tumor-suppressive function, EC109 cells without ing effect (Fig. 4C). Next, we also investigated whether Rab25 expression were stably transduced with lentivirus overexpression of Rab25 has a suppressive effect on angio- packaged with either a Rab25-expressing vector or an empty genesis. Human umbilical vein endothelial cells (HUVEC) vector (EV) control to generate Rab25-overexpressing were treated with conditioned media collected from Rab25 (EC109 Rab25) or control cells (EC109 EV). Similarly, KYSE30 overexpressed or suppressed cells and their respective con- cells, exhibiting high levels of Rab25 expression, were lenti- trols. Morphologic changes and differential tube-forming virally transduced with either Rab25 short hairpin RNA abilities of HUVECs following treatment with conditioned (shRNA; clones 849 and 852) or a non-target control (NTC) mediaareshowninFig.4D.HUVECsdisplayedelongated sequence to generate cells with Rab25 stably repressed tube-like structures following treatment with conditioned (KYSE30 shRNA 849 or 852) or control cells (KYSE30 NTC). media from Rab25 suppressed clones 849 and 852, as com- Stable Rab25 overexpression and knockdown were con- pared with control cell medium treatment. Conversely, firmed at both mRNA and protein levels by qPCR and treatment with conditioned media from Rab25-overex- Western blotting, respectively (Fig. 4A and B). Stable Rab25 pressed cells displayed a contrasting effect. These observa- overexpression significantly reduced the ability of the cells tions suggest a role for Rab25 in the regulation of invasion, to migrate and invade through Transwell chambers. Con- migration, and angiogenesis in ESCCs.

www.aacrjournals.org Cancer Res; 72(22) November 15, 2012 OF7

Tong et al.

Rab25 suppresses in vivo tumor formation and increased cell proliferation and microvessel density, respec- angiogenesis in ESCCs tively. Conversely, reduced PCNA and CD34 expression were An in vivo mouse model was further used to support our observed in KYSE30 NTC xenografts compared with Rab25- findings in vitro. Tumor formation ability of Rab25-overex- repressed tumors (Fig. 5B). pressed clone was significantly subdued as compared with EV control cells (Fig. 5A, left). On the contrary, mice injected with Rab25 drives ESCCs through a deregulated MAPK/ERK Rab25-repressed clone formed larger and more tumors than signaling pathway NTC cells (Fig. 5A, right). Serial sections from the xenograft Rab25 has previously been reported to control recycling of tumors were then subjected to hematoxylin and eosin (H&E) b1 integrin (11–13). Focal adhesion kinase (FAK), which loca- staining as well as IHC analysis. Histologic analysis and CK5/6 lizes with b-subunit of integrins, is the major and most IHC staining found that tumors formed with EC109 Rab25, extensively studied downstream player activated by integrins. EC109 EV, and KYSE30 shRNA 852 cells resembled an ESCC Activation and autophosphorylation of FAK has previously phenotype, whereas xenografts formed from KYSE30 NTC cells been shown to be critical in driving a number of cancer were composed mostly of necrotic cells and fibroblasts. Xeno- processes including promoting cell survival, cell proliferation, graft tissue sections were also examined for PCNA and CD34 and cell motility, through regulating downstream pathways expression by IHC. As compared with EC109 Rab25–overex- including the MAPK/ERK pathway (4, 22–24). In view of the pressing xenografts, xenografts generated with EV cells dis- importance of MAPK/ERK pathway in regulating cell invasive- played an enhanced PCNA and CD34 expression, indicative of ness and angiogenesis, we therefore examined whether Rab25

A EC109 EV (4/5) KYSE30 shRNA 852 (5/6)

EC109 Rab25 (2/5) KYSE30 NTC (3/6)

B EC109 EV EC109 Rab25 KYSE30 NTC KYSE30 shRNA 852

H&E (×100)

CK5/6 (×200)

300 *

200 **

100 PCNA (×200) 0 PCNA staining index EV Rab25 NTC 852

EC109 KYSE30 50 40 * ** 30 CD34 20 (×100) 10 0 Microvessel density Microvessel EV Rab25 NTC 852

EC109 KYSE30

Figure 5. Rab25 suppresses in vivo tumor initiation and angiogenesis in ESCCs. A, representative images of the subcutaneous tumors formed in nude mice following injection of Rab25-overexpressed or -repressed clones and their respective controls. Tumor incidence denoted in parentheses. B, H&E and IHC staining for CK5/6, PCNA, and CD34 expression in the resected xenografts. Bar chart summary of average staining intensity of PCNA and microvessel density in 5 randomly selected "hot spot" regions (, P < 0.05; , P 0.01).

OF8 Cancer Res; 72(22) November 15, 2012 Cancer Research

Rab25 in Esophageal Cancer

could regulate the expression of activated kinases in this investigating differential gene expression, chromosomal aber- particular signaling cascade. As compared with EV control, rations, and LOH using microarrays, comparative genomic Rab25 overexpression resulted in a reduction of phosphory- hybridization, and microsatellite DNA marker analysis tech- lated FAK and c-Raf with a concomitant decrease in the niques (25–28). In recent years, the advent of next-generation downstream phosphorylation of MEK1/2 and ERK (Fig. 6A). sequencing technologies has provided a new platform for Conversely, increased phosphorylation of FAK, c-Raf, MEK1/2, generating vast amounts of data at genomic, epigenomic, and ERK was detected in Rab25-suppressed cells, as compared transcriptomic, and proteomic levels by means of a variety of with its respective NTC (Fig. 6A). Interestingly, IHC staining in high-throughput technologies (29). In particular, the recent the resected xenografts likewise showed a reduced expression development of RNA sequencing has provided a new approach of phospho-ERK in Rab25-overexpressing xenografts. Con- for mapping and quantifying transcriptomes (30). Direct tab- versely, tumors formed with Rab25-repressed cells displayed ulation of the transcriptome by RNA-Seq has advantages over an enhanced expression of phospho-ERK (Fig. 6B). Taken existing technologies in several ways. First, unlike hybridiza- together, our results suggested that Rab25 exerts its tumor- tion-based approaches, RNA-Seq is not limited to detecting suppressive function in ESCCs through regulating the MAPK/ known genes but can also include the detection of novel ERK signaling pathway (Fig. 6C). transcripts and alternative splice forms (provided sequencing is conducted to a sufﬁcient depth). A second advantage of RNA- Discussion Seq relative to DNA microarrays is that RNA-Seq has high Past studies on the genetic alterations in ESCCs by our group sensitivity for detecting transcripts of low expressed genes and and other research laboratories have mainly been focused on that many sequencing reads can be unambiguously assigned to

A EC109 KYSE30 C EV Rab25 NTC 849 852

p-FAK β α 1.0 0.54 1.0 1.42 1.95 Total FAK 1.0 1.09 1.0 0.93 1.02 p-c-Raf (Ser259) 1.0 0.65 1.0 1.49 2.09 Total c-Raf Rab25 1.0 0.98 1.0 0.93 0.99 p-MEK1/2 (Ser217/221) FAK 1.0 0.66 1.0 2.14 1.76 P p-ERK (Thr202/Tyr204) 1.0 0.64 1.0 4.45 3.51 Total ERK P 1.0 0.99 1.0 0.93 0.85 β-Actin Raf 1.0 1.05 1.0 1.04 1.01

B P EC109 EV EC109 Rab25 200 P MEK 150 ** ** 100

50 P P 0 p-ERK staining index ERK KYSE30 NTC KYSE30 shRNA 852 EVRab25 NTC 852

p-ERK ( × 200) EC109 KYSE30

Migration Tumorigenesis Angiogenesis invasion

Figure 6. Rab25 drives ESCC tumorigenesis via a deregulated MAPK/ERK signaling pathway. A, detection of expression of phospho-FAK, total FAK, phospho- c-Raf (Ser259), total c-Raf, phospho-MEK1/2, phospho-ERK, and total ERK in Rab25-overexpressed or -repressed clones and their respective controls by Western Blotting. b-Actin was used as a loading control. B, phospho-ERK IHC staining in the resected xenografts generated from Rab25-overexpressed (EC109 Rab25) or -suppressed (KYSE30 shRNA 852) cells and their respective control cells. Bar chart summary of average staining intensity of phospho-ERK in 5 randomly selected "hot spot" regions (, P 0.01). C, a schematic diagram illustrating the proposed Rab25-regulated tumor-suppressive mechanism in ESCC tumorigenesis.

www.aacrjournals.org Cancer Res; 72(22) November 15, 2012 OF9

Tong et al.

the genome. Finally, it is also shown to have high levels of of Rab25 expression as a result of chromosome 1q22-23 reproducibility. RNA-Seq has now been successfully applied mutation has previously been suggested for breast cancer cell in the study of various disease models including Alzheimer lines (48), whereas upregulated Rab25 expression has been disease (31), leukemia (32), prostate (33, 34), and breast (35) suggested to be associated with chromosome amplification in cancers but not yet ESCCs. In this present study, RNA-Seq ovarian cancer (10) and PKA-dependent regulation of the was used to investigate differential gene expression in 12 proximal promoter in gastric cancer cell line model (21). Our nontumor and ESCC clinical samples. Using a stringent preliminary studies of DNA copy number in 2 ESCC cell lines statistical cutoff of P < 0.001, 1,730 commonly differentially (KYSE520 and EC109) and 2 ESCC clinical samples with no or expressed genes were identified. Pathway and Gene Ontog- low Rab25 expression suggested that downregulation of Rab25 eny analysis by DAVID in the 1,730 genes found 2 of the top 7 is not associated with deletion of the Rab25 gene (Supplemen- most enriched pathways to be associated with integrin tary Fig. S2). Although there is no solid explanation to con- signaling (i.e., integrin cell surface interactions and integrin clusively explain the different outcomes observed by different signaling pathway), suggesting that this pathway should play groups with regard to Rab25 expression, its regulation, its a critical role in driving ESCC tumorigenesis. Integrin sig- functional role, and its mediated pathways, several important naling has been extensively shown in the past to control factors should be considered. Most of the data that indicate initiation and progression of various cancer types. And this Rab25 as an oncogene is derived from studies in ovarian and is also true in the context of ESCCs, as a number of the breast cancers, which is in a completely different cellular identified differentially expressed genes involved in integrin context from ESCCs. The stage of the disease can also signaling in this study (listed in Supplementary Table S4) determine whether a certain gene acts as a tumor promoter have already been previously implicated in ESCC pathogen- or suppressor. For instance, TGF-b has been shown to act as esis. These include CAV1, ROCK1, ITGB1, ICAM1, VCL, PXN, a tumor suppressor in early stages of pancreatic cancer but and SHC1 (36–44). Here, we report the identification and then switches to become an oncogene in metastatic pan- characterization of a tumor suppressor gene Rab25 in creatic cancer (50). Furthermore, the cellular/mechanistic ESCCs, which was found to be significantly downregulated pathway most impacted may also be vital in determining the in the integrin signaling–associated pathway. nature of the action of Rab25 in different cellular models. For Rab25, with an epithelial distribution, was first discovered instance, Cheng and colleagues found Rab25 to exert tumor- with its expression enriched in the normal gastrointestinal suppressive properties in MDA-MB-231 breast cancer cells mucosa, kidney, and lung but absent in the brain, liver, and in part through effects on VEGF-A secretion and VEGFR-1 skeletal muscle (7). It is a complex molecule that has been expression (14), whereas our present study found Rab25 to described as an intracellular transport protein (45, 46). More exert tumor-suppressive effects in ESCCs through a deregu- recent studies have also shown that it is a tumor modulator as lated MAPK/ERK pathway. In ovarian and breast cancers, it can regulate tumor growth and angiogenesis. A handful of however, Rab25 has been shown to possess oncogenic func- studies have now been conducted to investigate the deregu- tions through suppressing Bak/Bax pro-apopototic mole- lated expression of Rab25 in different cancer types including cules and activation of PI3K/Akt pathway (10). These factors ovarian (10, 47), breast (14, 48), and colon cancers (13). On the may potentially contribute to determine whether Rab25 basis of these studies, Rab25 seems to display very contra- promotes or impedes tumor growth. Regardless, all these dicting roles dependent on cellular context. Concordant with data do suggest a multifunctional role of Rab25 in different findings in triple-negative breast and colon cancers (13, 14, 48), cellular contexts and either a gain or loss of Rab25 expres- we also found Rab25 to function as a tumor suppressor in sion could potentially lead to tumorigenesis in different ESCCs. And as Nam and colleagues reported for colon cancer organ models. (13), Rab25 was also found to have prognostic value in ESCCs. Cheng and colleagues have previously reported that Rab25 can Disclosure of Potential Conflicts of Interest fl act through multiple pathways to enhance apoptosis and to No potential con icts of interest were disclosed. suppress angiogenesis and invasion in triple-negative breast Authors' Contributions cancer by modulating VEGF-A and VEGFR-1 expression (14). Conception and design: M. Tong, K.W. Chan, S. Lok, X.-Y. Guan, S. Ma Contrary to their findings, our preliminary findings by qPCR Development of methodology: M. Tong, K.Y. Wong, J.-N. Chen show unaltered VEGF-A and VEGFR-1 expression in Rab25- Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): M. Tong, K.W. Chan, K.Y. Wong, J.-N. Chen, P.S. Kwan, overexpressed or -repressed ESCC cells (data not shown). K.H. Tang, Y.-R. Qin, S. Lok, S. Ma However, we did identify another pathway by which Rab25 Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): M. Tong, K.W. Chan, J.Y.J. Bao, S. Lok, S. Ma mediates tumorigenicity, metastasis, and angiogenesis in Writing, review and/or revision of the manuscript: M. Tong, K.W. Chan, J.Y. ESCC—through a deregulated FAK-Raf-MEK1/2-ERK signal- J. Bao, S. Lok, X.-Y. Guan, S. Ma ing. This study is also the first comprehensive study to identify Administrative, technical, or material support (i.e., reporting or orga- nizing data, constructing databases): K.W. Chan, P.S. Kwan, L. Fu, Y.-R. Qin, S a role of promoter hypermethylation in the inactivation of Lok, S. Ma Rab25 expression in ESCCs. It is interesting to note that Rab25 Study supervision: K.W. Chan, X.-Y. Guan, S. Ma has been suggested, along with a number of other genes (CDCA8, ATAD2, AURKA) to have a strong correlation for Acknowledgments The authors thank Dr. Kwai Wa Cheng of University of Texas MD Anderson methylation-dependent expression changes in ovarian cancer; Cancer Center (Houston, TX) for his generous gift of the monoclonal rabbit anti- although data are yet to be validated (49). On the contrary, loss human Rab25 antibody.

OF10 Cancer Res; 72(22) November 15, 2012 Cancer Research

Rab25 in Esophageal Cancer

Grant Support advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this This work was supported by a grant from the National Natural Science fact. Foundation of China (30971606) and the Sun Yat-Sen University "Hundred Talents Program" (85000–3171311). The costs of publication of this article were defrayed in part by the Received April 3, 2012; revised July 23, 2012; accepted September 5, 2012; payment of page charges. This article must therefore be hereby marked published OnlineFirst September 18, 2012.

References 1. Jemal A, Bray F, Center MM, Ferlay J, Ward E, Forman D. Global cancer 21. Xue H, Qiao Y, Ni P, Wang J, Chen C, Huang G. A CRE that binds CREB statistics. CA Cancer J Clin 2011;61:69–90. and contributes to PKA-dependent regulation of the proximal promot- 2. Li K. Mortality and incidence trends from esophagus cancer in selected er of human RAB25 gene. Int J Biochem Cell Biol 2011;43:348–57. geographic areas of China circa 1979–90. Int J Cancer 2002;102: 22. Renshaw MW, Price LS, Schwartz MA. Focal adhesion kinase med- 271–4. iates the integrin signaling requirement for growth factor activation of 3. Parise LV, Lee J, Juliano RL. New aspects of integrin signaling in MAP kinase. J Cell Biol 1999;147:611–8. cancer. Semin Cancer Biol 2000;10:407–14. 23. Sieg DJ, Hauck CR, Klingbeil CK, Schaefer E, Damsky CH, Schlaepfer 4. Hood JD, Cheresh DA. Role of integrins in cell invasion and migration. DD. FAK integrates growth-factor and integrin signals to promote cell Nat Rev Cancer 2002;2:91–100. migration. Nat Cell Biol 2000;2:249–56. 5. Schwartz MA, Ginsberg MH. Networks and crosstalk: integrin signal- 24. Dhillon AS, Hagan S, Rath O, Kolch W. MAP kinases signalling path- ling spreads. Nat Cell Biol 2002;4:65–8. ways in cancer. Oncogene 2007;26:3279–90. 6. Desgrosellier JS, Cheresh DA. Integrins in cancer: biological impli- 25. Kong KL, Kwong DL, Fu L, Chan TH, Chen L, Liu H, et al. Char- cations and therapeutic opportunities. Nat Rev Cancer 2010;10: acterization of a candidate tumor suppressor gene uroplakin 1A in 9–22. esophageal squamous cell carcinoma. Cancer Res 2010;70: 7. Goldenring JR, Shen KR, Vaughan HD, Modlin IM. Identification of a 8832–41. small GTP-binding protein, Rab25, expressed in the gastrointestinal 26. Qin YR, Fu L, Sham PC, Kwong DL, Zhu CL, Chu KK, et al. Single- mucosa, kidney, and lung. J Biol Chem 1993;25:18419–22. nucleotide polymorphism-mass array reveals commonly deleted 8. Casanova JE, Wang X, Kumar R, Bhartur SG, Navarre J, Woodrum JE, regions at 3p22 and 3p14.2 associate with poor clinical outcome et al. Association of Rab25 and Rab11a with the apical recycling in esophageal squamous cell carcinoma. Int J Cancer 2008;123: system of polarized Madin-Darby canine kidney cells. Mol Biol Cell 826–30. 1999;10:47–61. 27. Ueno T, Tangoku A, Yoshino S, Abe T, Toshimitsu H, Furuya T, et al. 9. Wang X, Kumar R, Navarre J, Casanova JE, Goldenring JR. Regulation Gain of 5p15 detected by comparative genomic hybridization as an of vesicle trafficking in Madin-Darby canine kidney cells by Rab11a and independent marker of poor prognosis in patients with esophageal Rab25. J Biol Chem 2000;275:29138–46. squamous cell carcinoma. Clin Cancer Res 2002;8:526–33. 10. Cheng KW, Lahad JP, Kuo WL, Lapuk A, Yamada K, Auersperg N, et al. 28. Li J, Liu Z, Wang Y, Yu Z, Wang M, Zhan Q, et al. Allelic imbalance of The RAB25 small GTPase determines aggressiveness of ovarian and chromosome 1q in esophageal squamous cell carcinomas from China: breast cancers. Nat Med 2004;10:1251–6. a novel region of allelic loss and significant association with differen- 11. Caswell PT, Spence HJ, Parsons M, White DP, Clark K, Cheng KW, tiation. Cancer Lett 2005;220:221–30. et al. Rab25 associates with alpha5beta1 integrin to promote invasive 29. Hawkins RD, Hon GC, Ren B. Next-generation genomics: an integra- migration in 3D microenvironments. Dev Cell 2007;13:496–510. tive approach. Nat Rev Genet 2010;11:476–86. 12. Dozynkiewicz MA, Jamieson NB, Macpherson I, Grindlay J, van den 30. Wang Z, Gerstein M, Snyder M. RNA-Seq: a revolutionary tool for Berghe PV, von Thun A, et al. Rab25 and CLIC3 collaborate to promote transcriptomics. Nat Rev Genet 2009;10:57–63. integrin recycling from late endosomes/lysosomes and drive cancer 31. Twine NA, Janitz K, Wilkins MR, Janitz M. Whole transcriptome progression. Dev Cell 2012;22:131–45. sequencing reveals gene expression and splicing differences in 13. Nam KT, Lee HJ, Smith JJ, Lapierre LA, Kamath VP, Chen X, et al. Loss brain regions affected by Alzheimer's disease. PLoS One 2011;6: of Rab25 promotes the development of intestinal neoplasia in mice and e16266. is associated with human colorectal adenocarcinomas. J Clin Invest 32. Wilhelm BT, Briau M, Austin P, Faubert A, Boucher G, Chagnon P, et al. 2010;120:840–9. RNA-seq analysis of 2 closely related leukemia clones that differ in their 14. Cheng JM, Volk L, Janaki DK, Vyakaranam S, Ran S, Rao KA. Tumor self-renewal capacity. Blood 2011;117:e27–38. suppressor function of Rab25 in triple-negative breast cancer. Int 33. Pflueger D, Terry S, Sboner A, Habegger L, Esqueva R, Lin PC, et al. J Cancer 2010;126:2799–812. Discovery of non-ETS gene fusions in human prostate cancer using 15. Xie D, Sham JS, Zeng WF, Lin HL, Che LH, Wu HX, et al. Heteroge- next-generation RNA sequencing. Genome Res 2011;21:56–67. neous expression and association of beta-catenin, p16 and c-myc in 34. Prensner JR, Iyer MK, Balbin OA, Dhanasekaran SM, Cao Q, Brenner multistage colorectal tumorigenesis and progression detected by JC, et al. Transcriptome sequencing across a prostate cancer cohort tissue microarray. Int J Cancer 2003;107:896–902. identifies PCAT-1, an unannotated lincRNA implicated in disease 16. Mortazavi A, Williams BA, McCue K, Schaeffer L, Wold B. Mapping and progression. Nat Biotechnol 2011;29:742–9. quantifying mammalian transcriptomes by RNA-Seq. Nat Methods 35. Edgren H, Murumagi A, Kangaspeska S, Nicorici D, Hongisto V, Kleivi 2008;5:621–8. K, et al. Identification of fusion genes in breast cancer by paired-end 17. Baggerly KA, Deng L, Morris JS, Aldaz CM. Differential expression in RNA-sequencing. Genome Biol 2011;12:R6. SAGE: accounting for normal between-library variation. Bioinformatics 36. Kato K, Hida Y, Miyamoto M, Hashida H, Shinohara T, Itoh T, et al. 2003;19:1477–83. Overexpression of caveolin-1 in esophageal squamous cell carcinoma 18. Shimada Y, Imamura M, Wagata T, Yamaguchi N, Tobe T. Character- correlates with lymph node metastasis and pathologic stage. Cancer ization of 21 newly established esophageal cancer cell lines. Cancer 2002;94:929–33. 1992;69:277–84. 37. Ando T, Ishiguro H, Kimura M, Mitsui A, Mori Y, Sugito N, et al. The 19. Tao Q, Huang H, Geiman TM, Lim CY, Fu L, Qiu GH, et al. Defective de overexpression of caveolin-1 and caveolin-2 correlates with a poor novo methylation of viral and cellular DNA sequences in ICF syndrome prognosis and tumor progression in esophageal squamous cell car- cells. Hum Mol Genet 2002;11:2091–102. cinoma. Oncol Rep 2007;18:601–9. 20. Weidner N, Semple JP, Welch WR, Folkman J. Tumor angiogenesis 38. Zhou J, Zhao LQ, Xiong MM, Wang XQ, Yang GR, Qiu ZL, et al. Gene and metastasis—correlation in invasive breast carcinoma. N Engl J expression profiles at different stages of human esophageal squa- Med 1991;324:1–8. mous cell carcinoma. World J Gastroenterol 2003;9:9–15.

www.aacrjournals.org Cancer Res; 72(22) November 15, 2012 OF11

Tong et al.

39. Fanucchi S, Veale RB. Delayed caspase-8 activation and enhanced 44. Bashir M, Kirmani D, Bhat HF, Baba RA, Hamza R, Naqash S, et al. integrin b1-activated FAK underpins anoikis in oesophageal P66shc and its downstream Eps8 and Rac1 proteins are upregulated in carcinoma cells harbouring mt p53-R175H. Cell Biol Int 2011;53: esophageal cancers. Cell Commun Signal 2010;8:13. 819–26. 45. Novick P, Zerial M. The diversity of Rab proteins in vesicle transport. 40. Rockett JC, Darnton SJ, Crocker J, Matthews HR, Morris AG. Expres- Curr Opin Cell Biol 1997;9:496–504. sion of HLA-ABC, HLA-DR and intercellular adhesion molecule-1 in 46. Stenmark H. Rab GTPases as coordinators of vesicle traffic. Nat Rev oesophageal carcinoma. J Clin Pathol 1995;48:539–44. Mol Cell Biol 2009;10:513–25. 41. Hosch SB, Izbicki JR, Piclmeier U, Stoecklein N, Niendorf A, Knoefel 47. Fan Y, Xin XY, Chen BL, Ma X. Knockdown of Rab25 expression by WT, et al. Expression and prognostic significance of immunoregulatory RNAi inhibits growth of human epithelial ovarian cancer cells in vitro molecules in esophageal cancer. Int J Cancer 1997;74:582–7. and in vivo. Pathology 2006;38:561–7. 42. Lifschitz-Mercer B, Czernobilsky B, Feldberg E, Geiger B. Expression 48. Cheng JM, Ding M, Aribi A, Shan P, Rao K. Loss of Rab25 expression in of the adherens junction protein vinculin in human basal and squamous breast cancer. Int J Cancer 2006;118:2957–64. cell tumors: relationship to invasiveness and metastatic potential. Hum 49. Wrzeszczynski KO, Varadan V, Byrnes J, Lum E, Kamalakaran S, Pathol 1997;28:1230–6. Levine DA, et al. Identification of tumor suppressors and oncogenes 43. Li BZ, Lei W, Zhang CY, Zhou F, Li N, Shi SS, et al. Increased from genomic and epigenetic features in ovarian cancer. PLoS One expression of paxillin is found in human oesophageal squamous 2011;6:e28503. cell carcinoma: a tissue microarray study. J Int Med Res 2008;36: 50. Truty MJ, Urrutia R. Basics of TGF-beta and pancreatic cancer. 273–8. Pancreatology 2007;7:423–35.

OF12 Cancer Res; 72(22) November 15, 2012 Cancer Research

Rab25 Is a Tumor Suppressor Gene with Antiangiogenic and Anti-Invasive Activities in Esophageal Squamous Cell Carcinoma

Man Tong, Kwok Wah Chan, Jessie Y.J. Bao, et al.

Cancer Res Published OnlineFirst September 18, 2012.

Updated version Access the most recent version of this article at: doi:10.1158/0008-5472.CAN-12-1269

Supplementary Access the most recent supplemental material at: Material http://cancerres.aacrjournals.org/content/suppl/2012/09/18/0008-5472.CAN-12-1269.DC1

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

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

Permissions To request permission to re-use all or part of this article, use this link http://cancerres.aacrjournals.org/content/early/2012/11/08/0008-5472.CAN-12-1269. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.