Human Cancer Biology

Dysregulated Expression of Stem Cell Factor Bmi1in Precancerous Lesions of the Gastrointestinal Tract KeisukeTateishi,1Miki Ohta,1Fumihiko Kanai,1, 2 Bayasi Guleng,1YasuoTanaka,1, 2 Yoshinari Asaoka,1 MotohisaTada,1Motoko Seto,1AmarsanaaJazag,1Lin Lianjie,1Makoto Okamoto,1Hiroyuki Isayama,1 MinoruTada,1HaruhikoYoshida,1Takao Kawabe,1and Masao Omata1

Abstract Purpose: It is important to identify the definitive molecular switches involved in the malignant transformation of premalignant tissues. is a specific characteristic of precan- cerous tissues, but not of cancers, which might reflect tumorigenesis-protecting mechanisms in premalignant lesions. Polycomb Bmi1, which is a potent negative regulator of the p16INK4 , suppresses senescence in primary cells and is overexpressed in various cancers. We hypothesized that Bmi1expression would also be dysregulated in precancerous lesions in human digestive precancerous tissues. Experimental Design: Bmi1 expression was investigated in cancerous and precancerous tissues of the digestive tract. The expression of , h-catenin, and Gli1 and the in vivo methyl- ation status of the p16 gene were also analyzed in serial sections of colonic precancerous lesions. Results: Bmi1was clearly overexpressed across a broad spectrum of gastrointestinal cancers, and the expression of Bmi1 increased in a manner that reflected the pathologic malignant features of precancerous colonic tissues (low-grade dysplasia, 12.9 F 2.0%; high-grade dyspla- sia, 82.9 F 1. 6 % ; c anc e r, 87. 5 F 2.4%). p16 was also strongly expressed in high-grade dyspla- sia, but not in cancers. p16 promoter methylation was detected only in some Bmi1-positive neoplastic cells. Conclusions: Bmi1overexpression was correlated with the malignant grades of human digestive precancerous tissues, which suggests that advanced Bmi1dysregulation might predict malignant progression. The abnormal Bmi1 expression might link to malignant transformation via the disturbance of orderly histone modification.

In cancer research, the mechanism of switching from prema- better understand carcinogenic processes. p16 and p14 encoded lignancy to established malignancy is currently the focus of by the Ink4a/Arf locus are important senescence biomarkers intensive investigation. Premalignant tumors often show (1). Aberrant mitogen-activated protein kinase signaling in oncogenic DNA changes that lead to abnormal proliferation, primary and immortalized fibroblasts induces the expression of whereas a large proportion of the cells in tumors undergo both of these markers (3, 4), which is a potent fail-safe -induced senescence. However, oncogenic cells bypass mechanism to prevent cells from engaging in uncontrolled this process due to the lack of oncogene-induced senescence proliferation. Indeed, in human malignancies, p16 is inacti- effectors, such as p16 or (1), and establish malignant vated at high frequency (5) via aberrant epigenetic changes or transformation. In other words, oncogene-induced senescence genomic deletions. Importantly, gene-targeting analysis has is a cellular response that might be crucial for protecting against shown that p16 suppression is indispensable for invasive cancer cancer development (1, 2). Therefore, it is important to formation (6). Combined, these findings suggest that inac- examine the mechanism of oncogene-induced senescence to tivation is the important step in oncogenesis and underline the significance of determining how expression inhibition is initiated in neoplastic lesions. Authors’ Affiliations: 1Department of Gastroenterology, University of Tokyo The epigenetic signatures of most cell types become fixed once and 2Department of Clinical Drug Evaluation, Graduate School of Medicine, The the cells differentiate or exit the . However, during University ofTokyo,Tokyo, Japan normal development, or in disease, some cells undergo major Received 2/27/06; revised 7/14/06; accepted 7/31/06. epigenetic ‘‘reprogramming’’ (7). Epigenetic alterations are The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance thought to be important factors for determining the character- with 18 U.S.C. Section 1734 solely to indicate this fact. istics of various cancer cells. For example, application to Requests for reprints: Keisuke Tateishi, Department of Gastroenterology, malignant cells of the ‘‘histone code’’ hypothesis, which posits Graduate School of Medicine, University of Tokyo, 7-3-1Hongo, Bunkyo-ku, Tokyo that different combinations of histone modifications mediate 113-8655, Japan. Phone: 81-3-3815-5411, ext. 33070; Fax: 81-3-3814-0021; E-mail: [email protected]. unique cellular responses, suggests that abnormal histone F 2006 American Association for Cancer Research. modifications in cancers serve as one of the molecular switches doi:10.1158/1078-0432.CCR-06-0449 for malignant transformation.

Clin Cancer Res 2006;12(23) December 1, 2006 6960 www.aacrjournals.org Downloaded from clincancerres.aacrjournals.org on September 25, 2021. © 2006 American Association for Cancer Research. Bmi1Overexpression in Precancerous Cells

Recently, polycomb group , which play pivotal roles Validated siRNAs were purchased from Dharmacon and transfected into in through chromatin modifications, have cells with Lipofectamine 2000 (Invitrogen, Carlsbad, CA) according been identified as (8). Given the role of polycomb to the protocol of the manufacturer. The knockdown efficiencies group in epigenetic modification, it is possible that were examined by immunoblotting 48 hours after treatment. After siRNA treatment for 48 hours, the bromodeoxyuridine (BrdUrd) uptake abnormal expression of polycomb group proteins contributes of the cells over the following 24 hours was determined with BrdUrd to the formation of cancer-specific cellular characteristics (8). ELISA (Roche, Basel, Switzerland). Three independent experiments were Bmi1 The polycomb group gene , which was originally iden- done. tified as an oncogene that induces B-cell or T-cell leukemia Methylation-specific PCR. The methylation status of p16 was (9, 10), is a potent negative regulator of the Ink4a/Arf locus determined by bisulfite treatment of the DNA, followed by methyla- (11–13). Bmi1 is overexpressed in several human cancers tion-specific PCR amplification, as described previously (23). In brief, (14, 15). Interestingly, given that Bmi1 protein suppresses the microdissected genomic DNA was denatured in 2 N NaOH at the senescence of primary cells (13) and is essential for 37jC for 10 minutes, and then incubated in 3 mol/L sodium bisulfite j maintaining stem cell fate (8, 16, 17), the overexpression of (pH 5.0) at 50 C for 16 hours in the dark. The following primer ¶ Bmi1might play a role in the escape from senescence in sequences were used: methylated sense, 5 -TTATTAGAGGGTGGGGCG- GATCGC-3¶; methylated antisense, 5¶-GACCCCGAACCGCGACCGTAA- premalignant cells. 3¶; unmethylated sense, 5¶-TTATTAGAGGGTGGGGTGGATTGT-3¶; and This work explored whether Bmi1plays a role in the unmethylated antisense, 5¶-CAACCCCAAACCACAACCATAA-3¶. malignant transformation of precancerous lesions of the Laser-capture microdissection. Microdissection of paraffin-embed- human digestive tract. ded tissue was done on H&E-stained slides. Genomic DNA was extracted from microdissected tissues as previously described (23). Each specimen was treated with lysis buffer that contained proteinase K and Materials and Methods incubated overnight at 56jC. The extracted DNA was purified with the QIAquick PCR Purification Kit (Qiagen, Valencia, CA). Cell lines and human tissue samples. Cell lines from four human Statistical analysis. The results are presented as the mean F SE. pancreatic carcinomas (SU8686, AsPC1, BxPC3, and CFPAC1), two Comparisons were made using Student’s two-tailed t test. P < 0.05 was gastric cancers (AGS and MKN7), three hepatic carcinomas (Alexander, considered statistically significant. Hep3B, and HepG2), a cholangiocarcinoma (HuCCT), and embryonic kidney (HEK293T) were purchased from the American Type Culture Collection (Manassas, VA) or the Japanese Riken Cell Bank (Tsukuba, Results Japan). Normal hepatocytes were purchased from Dainippon Pharma (Osaka, Japan) and normal diploid colonocytes (CCD841) were Bmi1 is accumulated in the cancerous cells of a broad spectrum purchased from American Type Culture Collection. Tissue specimens of gastrointestinal tumors. In normal gastrointestinal tissues, obtained during surgery or endoscopic resection were obtained from the levels of Bmi1mRNA expression vary with the tissue; e.g., the University of Tokyo Hospital (Tokyo, Japan) and Motojima Bmi1mRNA is detected in the pancreas but not in the small or Memorial Hospital (Gumma, Japan) after approval from the medical large intestines (24). We analyzed the expression in various ethics committee and the acquisition of informed consent. Some human digestive tract tumor tissues, including pancreas (Fig. 1A samples of colonic cancer and surrounding precancerous tissue were and B), bile duct (Fig. 1C and D), stomach (Fig. 1E and F), and collected from identical specimens. Formalin-fixed, paraffin-embedded colon (Fig. 1G and H) cancers. Bmi1 accumulation was sections were examined by H&E staining and immunohistochemistry. Plasmids, antibodies, and reagents. The Gli1expression plasmid detected in most cancerous cells, whereas tumor stromal tissues pcDNA3/Gli1 (18, 19) and empty control plasmids were used. The Smo and normal epithelial cells did not express this protein (Fig. 1E inhibitor cyclopamine (Toronto Research Chemicals, North York, and G), in contrast to bronchial epithelium, which shows Ontario, Canada; refs. 20, 21) was dissolved in DMSO. For widespread expression of Bmi1(25). Similarly, in the cancer immunoblotting, mouse anti-Bmi1(1:500;Upstate Biotechnology, cell lines, Bmi1was expressed in all 10cell lines derived from Lake Placid, NY), mouse anti-p16 (JC2; 1:1000; Lab Vision, Westing- stomach, liver, pancreas, and bile duct cancers, but weakly in house, CA), and mouse anti-h-actin (1:5,000; Sigma-Aldrich, St. Louis, normal cells (Fig. 1I). These results indicate the existence of MO) antibodies were used as the primary labeling antibodies, and cancer-specific induction mechanisms for Bmi1in various appropriate horseradish peroxidase–conjugated antibodies (1:2,000; cancers. Amersham, Uppsala, Sweden) were used as the secondary antibodies. Bmi1 shows a gradient of expression in human colonic The enhanced chemiluminescence-plus system (Amersham Bioscien- ces) was used for detection. The siRNA for Bmi1was purchased from neoplasms. To evaluate Bmi1expression in the precancerous Dharmacon (Lafayette, CO). lesions of digestive tissues, we examined the expression status Immunohistochemistry. Immunohistochemistry was done with in human colonic neoplastic lesions. Because colonic neoplasia mouse anti-Bmi1 (1:100), mouse anti-p16 (1:1,000), mouse anti-h- is the most intensively studied of the multistep carcinogenesis catenin (1:50; Transduction Labs, San Jose, CA), and rabbit anti-Gli1 models, it offers an excellent setting in which to examine (1:50; Chemicon, Temecula, CA) antibodies, together with the the status of Bmi1in human precancerous lesions. The appropriate biotinylated secondary antibodies (1:500), as previously tumors were pathologically classified into three groups depend- described (20, 22). Endogenous peroxidase activity was blocked with ing on the malignant grade: low-grade intraepithelial dysplasia, 3% H2O2 for 10 minutes at room temperature. Antigen was retrieved by high-grade intraepithelial dysplasia, and cancer. Interestingly, boiling the tissue in 0.01mol/L sodium citrate (pH 6.0) for 10minutes. the precancerous lesions showed distinct Bmi1expression Proteins were visualized using the standard 3,3¶-diaminobenzidine protocol. The percentages of Bmi1- and p16-positive cells in the glands patterns relative to the pathologic grade of neoplasia (Fig. 2A were calculated based on the staining of 50 randomly selected glands and B). As shown in Fig. 2C, even in a single gland, cells from each tissue examined by microscopy. of different atypical grades showed distinct patterns. We scored Cell proliferation assay. The Bmi1-siRNA cocktail or control reagent the percentages of Bmi1-positive cells by gland with different was added to some of the wells at the start of the experiment (0 hours). atypical grades (n = 50 glands per group). In low-grade

www.aacrjournals.org 6961 Clin Cancer Res 2006;12(23) December 1, 2006 Downloaded from clincancerres.aacrjournals.org on September 25, 2021. © 2006 American Association for Cancer Research. Human Cancer Biology

lines in our experiments (Figs. 1I and 3A). Similarly, p16 was not detected in a large section of the normal colon or normal cell lines (ref. 27, and data not shown; Fig. 1I). However, p16 was clearly expressed in colonic precancerous lesions as previously reported (27), although the distribution was not uniform, even in identical tumors (Fig. 3B). We classified the distribution of p16-positive cells according to atypical grade, using the same tumors, with Bmi1classifica- tion (Fig. 3C). Numerous p16-positive cells were noted in the high-grade intraepithelial dysplasia (14.3 F 4.9%) compared with cancers (2.0 F 1.4%; Fig. 3C). Given the

Fig. 1. Bmi1expression in digestive tissue samples. Immunohistochemical Bmi1 staining of pancreas (A and B), bile duct (C and D), stomach (F), and colon (H) cancer tissues was done, as well as in normal stomach (E)andcolon(G) tissues. All of the sections are counterstained with hematoxylin. Magnification, Â100 (A-D); Â200 (E-H). I, immunoblotting for Bmi1and p16 in digestive tract cancer and normal cell lines. intraepithelial dysplasia, in which the nuclei were limited primarily to the lower halves of the atypical cells in the glands, the proportion of Bmi1-positive cells was 12.9 F 2.0% (Fig. 2D). Conversely, for high-grade intraepithelial dysplasia, in which the nuclei were translocated to higher levels, Bmi1was detected in 82.9 F 1.6% of the cells, a frequency that is similar to that of cancerous cells (87.5 F 2.4%; Fig. 2D). These findings indicate that Bmi1expression is connected to the mechanism that determines malignant potential. The p16 protein is expressed in precancerous colonic tissues, but not in the cancers. The Ink4a/Arf locus is a target of the polycomb complex, which includes Bmi1. Previous studies Fig. 2. Distinct expression patterns of Bmi1in precancerous lesions of the colon. noted p16 deficiency in many cancers, and a distinct H&E staining (A) and Bmi1immunostaining (B and C) of precancerous colon distribution of Bmi1with p16has been observed in colon specimens (magnification, Â200). D, quantitative analysis of Bmi1expression in colon tumors. The results are shown as scatter plots of 50 glands derived from 12 cancers (26). The expression of p16 was not detected in specimens. Low, low-grade intraepithelial neoplasia; High, high-grade intraepithelial either most of the colon cancer tissues or certain cancer cell neoplasia. Columns, mean; bars, SE. *, P < 0.05; n.s., not significant.

Clin Cancer Res 2006;12(23) December 1, 2006 6962 www.aacrjournals.org Downloaded from clincancerres.aacrjournals.org on September 25, 2021. © 2006 American Association for Cancer Research. Bmi1Overexpression in Precancerous Cells

promoter methylation is not always induced by Bmi1over- expression. In other words, Bmi1might have a mechanism of p16 suppression that is independent of the promoter methyl- ation during the transition from p16-positive precancerous tissues to p16-negative cancers. The aberrant expression of Bmi1 is independent of Wnt or Hedgehog signaling in neoplastic gastrointestinal tissues. The induction mechanisms of Bmi1in neoplastic cells have not been fully elucidated. The Wnt and Hedgehog (Hh) signaling pathways are important for the normal development of gastrointestinal tissues (20, 28–31), and it has been spe- culated that they are related to oncogenesis in gastrointes- tinal tumors (20, 29). To unravel the relationship between Wnt signaling and Bmi1expression, the expression patterns of h-catenin and Bmi1were compared in serial sections of precancerous colonic lesions (Fig. 5A and B). As shown in Fig. 5B, glands in which h-catenin was overexpressed in the cytoplasm (arrow) were Bmi1positive, although the Bmi1- positive cells did not always show h-catenin accumulation (arrowhead). Because Bmi1has been reported as being downstream of Hh signaling in cerebellar granule cells (32), we examined whether Hh signaling induces Bmi1in cancer cells. As a result, neither Gli1expression nor nuclear accumulation of Gli1was significant, irrespective of the expression of Bmi1in the colonic neoplasms (Fig. 5A and C). In addition, Hh signaling is activated in the upper gastrointestinal cancer cells (20, 29) and we reported that constitutive Hh expression activated the signaling in the human gastric cancer AGS cells (33). In the AGS cells, Gli1 overexpression did not induce Bmi1up-regulation (Fig. 5D). Moreover, when the signaling cascade in the AGS cells was blocked by treatment with the Smo inhibitor cyclopamine, Bmi1expression was not affected (Fig. 5D). Combined, these results suggest that Bmi1expression is independent of Hh or Wnt signaling, and that different induction mechanisms exist in gastrointestinal neoplastic cells. Bmi1 protein is not essential for DNA replication in cancer Fig. 3. P16 expression patterns in neoplastic lesions of the colon. Colon cancers (A) and precancerous tissues (B) were immunostained for Bmi1and p16 cells. Cyclin-dependent kinase inhibitor p16 leads to cell [magnification, Â100 (A); Â200 (B)]. Arrow, p16-positive region; arrowhead, cycle arrest by affecting the retinoblastoma protein (34, 35). p16-negative region. C, quantitative analysis of p16 expression in colon tumors. As Bmi1is a potent negative regulator of p16and has The results are shown as scatter plots of 50 glands derived from 12 specimens. Columns, mean; bars, SE. *, P < 0.05; n.s., not significant. dissociation of the expression patterns in some cancer cells (26), it seems that a significant change in p16 expression is induced in Bmi1and p16double-positive cells with high- grade intraepithelial dysplasia, which ultimately leads to malignant transformation. Methylation of the p16 gene promoter occurs in a part of Bmi1- positive cells of precancerous colonic tissues. Given the histone modification functions of polycomb protein Bmi1, aberrant epigenetic alterations were suspected in the cells overexpressing Bmi1. To determine whether dysregulated Bmi1 expression coexists with aberrant DNA methylation, the methylation status of the CpG island in the p16INK4a gene promoter was examined in the colonic tumors. The genomic DNA was selectively isolated from the neoplastic portions of four tumor Fig. 4. Analysis of p16 promoter methylation in colonic neoplastic lesions. tissues that had a distinct status for the Bmi1and p16proteins A, representative image of laser-capture microdissection of a colonic adenoma (Fig. 4A). In the Bmi1-positive glands, p16 methylation was (arrow, p16-positive region; arrowhead, p16-negative region). B, representative image of methylation-specific PCR analysis of the p16 promoter lesion in detected in only some of the Bmi1and p16double-positive microdissected tissues. U, unmethylated; M, methylated. Bmi1+, Bmi1expression precancerous lesions (Fig. 4B), which indicates that p16 positive; p16+, p16 expression positive.

www.aacrjournals.org 6963 Clin Cancer Res 2006;12(23) December 1, 2006 Downloaded from clincancerres.aacrjournals.org on September 25, 2021. © 2006 American Association for Cancer Research. Human Cancer Biology

Discussion

The global histone modification patterns have been described as being responsible for the heterogeneity of tumors and have been correlated with clinical behavior (36, 37). A recent report has documented that the Bmi1-associated gene signature is a powerful predictor of a short interval to distant metastasis and poor survival after therapy (36). Combined, the various aberrant histone modifications may contribute to the characteristics of cancer. Indeed, previous reports have documented that some histone modifiers are overexpressed in cancers. For example, high-level enhancer of zeste homologue 2 (EZH2) expression was localized to the more primitive malignant cell types, often in combination with high-level Bmi1expression (38). The functional relationships between abnormal enzyme-based histone modification and oncogene- induced signaling in cancers need to be clarified. Polycomb group protein Bmi1can be recruited to histone H3 at Lys27 methylated by EZH2 (39–41). Both EZH2 and Bmi1are up-regulated in many cancers (42–45). Altered epigenetic modifications caused by the dysregulated expression of EZH2 or Bmi1might contribute to tumor progression from precancerous cells. A recent report has documented that EZH2 interacts with DNA methyltransferases and is associated with methyltransferase activities (46). Although our findings do not exclude situations in which Bmi1overexpression cooperates with P16 promoter methylation, abnormal histone modifica- tions could silence gene expression without the involvement of DNA methylation (47–49). Although it remains to be determined whether the overexpression of Bmi1and EZH2 induces p16 gene suppression in precancerous cells, the mechanism linking the dysregulated histone modification to the epigenetic aspect of oncogenesis should be investigated further.

Fig. 5. Bmi1expression in cancer cells is independent of Wnt and Hh signaling. The colonic precancerous specimens were stained for Bmi1 (A), h-catenin (B), and Gli1 (C). D, AGS cells were transfected with pCMV-Gli1or control vector, and the effects on Bmi1expression were analyzed by immunoblotting (left). The AGS cells were treated with cyclopamine (right).

an essential role in regulating the proliferative activities of normal and leukemic stem cells (16), we examined whether the Bmi1protein is a positive cell cycle regulator in gastric cancer cells. Because Bmi1expression was diminished 48 hours after siRNA treatment in AGS cells (Fig. 6A), the in vitro BrdUrd uptake assay was done after siRNA treatment for 24 hours. As a result, BrdUrd uptake was not affected by the knockdown of Bmi1(Fig. 6B), which indicates that Bmi1is not essential for the growth of AGS cells. Because Bmi1is already overexpressed in precancerous lesions, the growth advantage with Bmi1overexpression might be given full play in premalignant cells, rather than in established cancer cells. Previously, we reported that the knockdown of Gli1leads to growth retardation of AGS cells (33). This result seems to be consistent with the finding that Bmi1does Fig. 6. Bmi1does not regulate the in vitro growth of gastric cancer cells. A, Bmi1 knockdown by siRNA. B, AGS cells were treated for 48 hours with siRNA for Bmi1 not always mediate the Hh signaling activity in gastric cancer or the control, and the BrdUrd assay was done 24 hours later.Three independent cells (Fig. 5). experiments were done. n.s., not significant.

Clin Cancer Res 2006;12(23) December 1, 2006 6964 www.aacrjournals.org Downloaded from clincancerres.aacrjournals.org on September 25, 2021. © 2006 American Association for Cancer Research. Bmi1Overexpression in Precancerous Cells

Given that our data do not show a perfect correlation brain tumors showed high-level expression of Bmi1among between Bmi1 and p16 expression (Figs. 1I and 3), it is possible various stem cell markers (51). It will be important to address that Bmi1functions through downstream molecules other than whether Bmi1induces the stem cell characteristics of cancer p16 in the tumors. In other words, the Bmi1-driven pathway cells. might have various significant roles other than cell proliferation that benefit through p16 suppression in neoplastic cells (50). For example, Bmi1might be linked to chemotherapy resistance Acknowledgments (36) or to stem cell-like properties (17, 32). Recently, cancer We thank Mitsuko Tsubouchi and Sanae Ogawa for technical assistance; stem cells were also found to reside within solid tumors, Drs. T. Motojima and S. Yamada (Division of Abdominal Surgery, Motojima General including several types of brain cancer (51, 52) and breast Hospital, Gumma, Japan), Dr. T. Oyama (Division of Pathology, Motojima General carcinomas (53). If stem cells are characterized by chemother- Hospital), Dr. H. Sasaki (RIKEN Center for Developmental Biology, Kobe, Japan), and Drs. B. Vogelstein and K.W. Kinzler (The Howard Hughes Medical Institute, apy resistance (54), Bmi1might play a role in the stem cell Sidney Kimmel Comprehensive Cancer Center and Program in Human Genetics properties of cancer cells, as occurs in other stem cell lineages. and Molecular Biology, Johns Hopkins Medical Institutions, Baltimore, MD) for Indeed, cancer stem cells cultured from a panel of pediatric supplying key materials.

References 1. Collado M, Gil J, Efeyan A, et al. Tumour biology: se- INK4A-ARF locus expression. Br J Cancer 2001;84: 32. Leung C, Lingbeek M, Shakhova O, et al. Bmi1is nescence in premalignant tumours. Nature 2005;436: 1372 ^ 6 . essential for cerebellar development and is overex- 642. 16. Lessard J, Sauvageau G. Bmi-1determines the pro- pressed in human medulloblastomas. Nature 2004; 2. Braig M, Lee S, Loddenkemper C, et al. Oncogene- liferative capacity of normal and leukaemic stem cells. 428:337^ 41. induced senescence as an initial barrier in lymphoma Nature 2003;423:255^ 60. 33. Ohta M,Tateishi K, Kanai F, et al. p53-Independent development. Nature 2005;436:660 ^ 5. 17. ParkIK,QianD,KielM,etal.Bmi-1isrequiredfor negative regulation of p21/cyclin-dependent kinase- 3. Serrano M, Gomez-Lahoz E, DePinho RA, Beach D, maintenance of adult self-renewing haematopoietic interacting protein 1 by the sonic hedgehog-glioma- Bar-Sagi D. Inhibition of ras-induced proliferation and stem cells. Nature 2003;423:302^ 5. associated oncogene 1 pathway in gastric carcinoma cellular transformation by p16INK4. Science1995;267: 18. Kinzler KW, Ruppert JM, Bigner SH, Vogelstein B. cells. Cancer Res 2005;65:10822^9. 249 ^ 52. The GLI gene is a member of the Kruppel family of zinc 34. Sharpless NE, DePinho RA. The INK4A/ARF locus 4. Serrano M, Lin AW, McCurrach ME, Beach D, Lowe finger proteins. Nature 1988;332:371 ^ 4. and its two gene products. Curr Opin Genet Dev1999; SW. Oncogenic ras provokes premature cell senes- 19. Sasaki H, Hui C, Nakafuku M, Kondoh H. A bind- 9:22^30. cence associated with accumulation of p53 and ing site for Gli proteins is essential for HNF-3h floor 35. Lowe SW, Sherr CJ. Tumor suppression by Ink4a- p16INK4a. Cell 1997;88:593^ 602. plate enhancer activity in transgenics and can re- Arf:progress and puzzles. Curr Opin Genet Dev 5. Baylin SB, Herman JG, Graff JR, Vertino PM, Issa spond to Shh in vitro.Development1997;124: 2003;13:77^ 83. JP. Alterations in DNA methylation: a fundamental 1313 ^ 2 2 . 36. Glinsky GV, Berezovska O, Glinskii AB. Microar- aspect of neoplasia. Adv Cancer Res 1998;72: 20.Thayer SP,di Magliano MP,Heiser PW, et al. Hedge- ray analysis identifies a death-from-cancer signa- 141 ^ 9 6. hog is an early and late mediator of ture predicting therapy failure in patients with 6. Aguirre AJ, Bardeesy N, Sinha M, et al. Activated tumorigenesis. Nature 2003;425:851 ^ 6. multiple types of cancer. J Clin Invest 2005;115: Kras and Ink4a/Arf deficiency cooperate to produce 21. Taipale J, Cooper MK, Maiti T, Beachy PA. Patched 15 0 3 ^ 21. metastatic pancreatic ductal adenocarcinoma. Genes acts catalytically to suppress the activity of Smooth- 37. Seligson DB, Horvath S, Shi T, et al. Global histone Dev 2003;17:3112 ^ 26. ened. Nature 2002;418:892 ^7. modification patterns predict risk of 7. Dean W, Lucifero D, Santos F. DNA methylation in 22. Tateishi K, Omata M,Tanaka K, ChibaT.The NEDD8 recurrence. Nature 2005;435:1262 ^ 6. mammalian development and disease. Birth Defects system is essential for cell cycle progression and mor- 38. van Kemenade FJ, Raaphorst FM, Blokzijl T, et al. Res Part C EmbryoToday 2005;75:98 ^ 111. phogenetic pathway in mice. J Cell Biol 2001;155: Coexpression of BMI-1 and EZH2 polycomb-group 8.Valk-Lingbeek ME, Bruggeman SW, van Lohuizen M. 571 ^ 9. proteins is associated with cycling cells and degree Stem cells and cancer; the polycomb connection. Cell 23. Moskaluk CA, Kern SE. Microdissection and poly- of malignancy in B-cell non-Hodgkin lymphoma. 2004;118:409 ^ 18. merase chain reaction amplification of genomic DNA Blood 2001;97:3896 ^ 901. 9. Haupt Y, Alexander WS, Barri G, Klinken SP, from histological tissue sections. Am J Pathol 1997; 39. Cao R, Wang L, Wang H, et al. Role of histone H3 Adams JM. Novel zinc finger gene implicated as 150:1547^52. lysine 27 methylation in Polycomb-group silencing. myc collaborator by retrovirally accelerated lympho- 24. Gunster MJ, Raaphorst FM, Hamer KM, et al.Differ- Science 2002;298:1039 ^ 43. magenesis in E mu-myc transgenic mice. Cell 1991; ential expression of human Polycomb group proteins 40. Czermin B, Melfi R, McCabe D, Seitz V, Imhof A, 65:753^63. in various tissues and cell types. J Cell Biochem Pirrotta V. Drosophila enhancer of Zeste/ESC com- 10. vanLohuizenM,VerbeekS,ScheijenB, 2001;81:129^ 43. plexes have a histone H3 methyltransferase activity Wientjens E, van der Gulden H, Berns A. Identifi- 25. Breuer RH, Snijders PJ, Smit EF, et al. Increased that marks chromosomal Polycomb sites. Cell 2002; cation of cooperating oncogenes in E mu-myc expression of the EZH2 polycomb group gene in 111: 18 5 ^ 9 6 . transgenic mice by provirus tagging. Cell 1991;65: BMI-1-positive neoplastic cells during bronchial carci- 41. Kuzmichev A, Nishioka K, Erdjument-Bromage H, 737 ^ 52. nogenesis. Neoplasia 2004;6:736 ^ 43. Tempst P, Reinberg D. Histone methyltransferase 11. Jacobs JJ, Scheijen B, Voncken JW, Kieboom K, 26. Kim JH,Yoon SY, Kim CN, et al. The Bmi-1 oncopro- activity associated with a human multiprotein complex Berns A, van Lohuizen M. Bmi-1 collaborates with tein is overexpressed in human and containing the Enhancer of Zeste protein. Genes Dev c-Myc in tumorigenesis by inhibiting c-Myc-induced correlates with the reduced p16INK4a/p14ARF pro- 2002;16:2893 ^ 905. apoptosis via INK4a/ARF. Genes Dev 1999;13: teins. Cancer Lett 2004;203:217^ 24. 42. Visser HP, Gunster MJ, Kluin-Nelemans HC, et al. 2678 ^ 90. 27. Dai CY, Furth EE, Mick R, et al. p16(INK4a) expres- The Polycomb group protein EZH2 is up-regulated in 12. Dimri GP, Martinez JL, Jacobs JJ, et al. The Bmi-1 sion begins early in human colon neoplasia and corre- proliferating, cultured human mantle cell lymphoma. oncogene induces telomerase activity and immortal- lates inversely with markers of cell proliferation. BrJ Haematol 2001;112:950 ^ 8. izes human mammary epithelial cells. Cancer Res Gastroenterology 2000;119:929 ^42. 43.Varambally S, Dhanasekaran SM, Zhou M, et al. 2002;62:4736 ^ 45. 28. Kinzler KW, Vogelstein B. Lessons from hereditary The polycomb group protein EZH2 is involved in 13. Itahana K, Zou Y, Itahana Y, et al. Control of the rep- colorectal cancer. Cell 1996;87:159 ^ 70. progression of prostate cancer. Nature 2002;419: licative life span of human fibroblasts by p16 and the 29. Berman DM, Karhadkar SS, Maitra A, et al. Wide- 624 ^ 9. polycomb protein Bmi-1. Mol Cell Biol 2003;23: spread requirement for Hedgehog ligand stimulation 44. Bracken AP,Pasini D, Capra M, Prosperini E, Colli E, 389^401. in growth of digestive tract tumours. Nature 2003; Helin K. EZH2 is downstream of the pRB-E2F path- 14. BeaS,TortF,PinyolM,etal.BMI-1geneamplifica- 425:846 ^ 51. way, essential for proliferation and amplified in cancer. tion and overexpression in hematological malignan- 30. Taipale J, Beachy PA. The Hedgehog and Wnt EMBO J 2003;22:5323^ 35. cies occur mainly in mantle cell lymphomas. Cancer signalling pathways in cancer. Nature 2001;411: 45. Kleer CG, Cao Q, Varambally S, et al. EZH2 is a Res 2001;61:2409 ^ 12. 349^54. marker of aggressive breast cancer and promotes neo- 15. Vonlanthen S, Heighway J, Altermatt HJ, et al. 31. Pasca di Magliano M, Hebrok M. Hedgehog signal- plastic transformation of breast epithelial cells. Proc The bmi-1 oncoprotein is differentially expressed in ling in cancer formation and maintenance. Nat Rev Natl Acad Sci US A 2003;100:11606^11. non-small cell lung cancer and correlates with Cancer 2003;3:903 ^ 11. 46.Vire E, Brenner C, Deplus R, et al. The Polycomb

www.aacrjournals.org 6965 Clin Cancer Res 2006;12(23) December 1, 2006 Downloaded from clincancerres.aacrjournals.org on September 25, 2021. © 2006 American Association for Cancer Research. Human Cancer Biology

group protein EZH2 directly controls DNA methyla- 49. Bachman KE, Park BH, Rhee I, et al. Histone ous stem cells can arise from pediatric brain tumors. tion. Nature 2006;439:871 ^ 4. modifications and silencing prior to DNA methyla- Proc Natl Acad Sci U S A 2003;100:15178 ^ 83. 47. Lewis A, Mitsuya K, Umlauf D, et al. Imprinting on tion of a tumor suppressor gene. Cancer Cell 52. Singh SK, Clarke ID,Terasaki M, et al. Identification distal 7 in the placenta involves repres- 2003;3:89 ^ 95. of a in human brain tumors. Cancer sive histone methylation independent of DNA methyl- 50. Molofsky AV, He S, Bydon M, Morrison SJ, Res 2003;63:5821 ^ 8. ation. Nat Genet 2004;36:1291 ^ 5. Pardal R. Bmi-1 promotes neural stem cell self- 53. Al-Hajj M, Wicha MS, Benito-Hernandez A, 48. Umlauf D, Goto Y, Cao R, et al. Imprinting along renewal and neural development but not mouse Morrison SJ, Clarke MF. Prospective identification the Kcnq1 domain on mouse chromosome 7 involves growth and survival by repressing the p16Ink4a of tumorigenic breast cancer cells. Proc Natl Acad repressive histone methylation and recruitment of and p19Arf senescence pathways. Genes Dev Sci U S A 2003;100:3983 ^ 8. Polycomb group complexes. Nat Genet 2004;36: 2005;19:1432^ 7. 54. Dean M, FojoT, Bates S.Tumour stem cells and drug 12 9 6 ^ 3 0 0. 51. Hemmati HD, Nakano I, Lazareff JA, et al. Cancer- resistance. Nat Rev Cancer 2005;5:275 ^ 84.

Clin Cancer Res 2006;12(23) December 1, 2006 6966 www.aacrjournals.org Downloaded from clincancerres.aacrjournals.org on September 25, 2021. © 2006 American Association for Cancer Research. Dysregulated Expression of Stem Cell Factor Bmi1 in Precancerous Lesions of the Gastrointestinal Tract

Keisuke Tateishi, Miki Ohta, Fumihiko Kanai, et al.

Clin Cancer Res 2006;12:6960-6966.

Updated version Access the most recent version of this article at: http://clincancerres.aacrjournals.org/content/12/23/6960

Cited articles This article cites 54 articles, 18 of which you can access for free at: http://clincancerres.aacrjournals.org/content/12/23/6960.full#ref-list-1

Citing articles This article has been cited by 5 HighWire-hosted articles. Access the articles at: http://clincancerres.aacrjournals.org/content/12/23/6960.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://clincancerres.aacrjournals.org/content/12/23/6960. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.

Downloaded from clincancerres.aacrjournals.org on September 25, 2021. © 2006 American Association for Cancer Research.