Oncogene (2010) 29, 2605–2615 & 2010 Macmillan Publishers Limited All rights reserved 0950-9232/10 $32.00 www.nature.com/onc REVIEW RUNX3 is multifunctional in carcinogenesis of multiple solid tumors

LSH Chuang1 and Y Ito1,2,3

1Cancer Biology Program, Cancer Science Institute of Singapore, National University of Singapore, Singapore; 2Department of Medicine, Yong Loo Lin School of Medicine, National University of Singapore, Singapore and 3Institute of Molecular and Cell Biology, Singapore

The study of RUNX3 in tumor pathogenesis is a rapidly Sgpp1, Ncam1, p21WAF1) (Wotton et al., 2008), the expanding area of cancer research. Functional inactiva- RUNX have been shown to perform distinct tion of RUNX3—through mutation, epigenetic silencing, functions that are dependent on tissue type (Brady and or cytoplasmic mislocalization—is frequently observed in Farrell, 2009). This review explores the dualistic solid tumors of diverse origins. This alone indicates that associations of RUNX3 with cancer with emphasis on RUNX3 inactivation is a major risk factor in tumorigen- the emerging concept of RUNX3 as a multifunctional esis and that it occurs early during progression to suppressor of solid tumors. malignancy. Conversely, RUNX3 has also been described to have an oncogenic function in a subset of tumors. Although the mechanism of how RUNX3 switches from RUNX3 inactivation is prevalent in solid tumors tumor suppressive to oncogenic activity is unclear, this is of clinical relevance with implications for cancer detection Hemizygous deletion of RUNX3 and prognosis. Recent developments have significantly The location of RUNX3 at 1p36, a deletion contributed to our understanding of the pleiotropic tumor hotspot in diverse cancers of epithelial, hematopoietic, suppressive properties of RUNX3 that regulate major and neural origins (Bagchi and Mills, 2008), prompts signaling pathways. This review summarizes the important the following questions: How often is RUNX3 deleted? findings that link RUNX3 to tumor suppression. Does RUNX3 deletion confer any cancer growth Oncogene (2010) 29, 2605–2615; doi:10.1038/onc.2010.88; advantage? Examination of gastric cancer tissue speci- published online 29 March 2010 mens revealed hemizygous deletions of RUNX3 at Keywords: RUNX3; tumor suppressor; solid tumors; frequencies which increased with tumor grade (Li et al., functional inactivation; TGF-b; Wnt 2002). Likewise, RUNX3 hemizygous deletion and concomitant decrease in RUNX3 expression are also prevalent in gastric, bile duct, pancreatic, and lung cancer cell lines (Li et al., 2002; Wada et al., 2004; Introduction Yanada et al., 2005). This suggests that downregulation of RUNX3 expression is an early occurrence during RUNX family of transcription factors progression to malignancy, and thus, provides a strong The Runt-related transcription factors (RUNX) describe hint of RUNX3 tumor suppressive function. Moreover, a family of evolutionarily conserved metazoan proteins, it is noteworthy that mapping to human 1p36 which share the highly homologous N-terminal Runt show syntenic conservation with mouse chromosome 4, domain, a DNA-binding, and heterodimerization region lending weight to the theory that this region harbors a of about 120 amino acids. Initially discovered as a cluster of tumor suppressor genes, which work coopera- heterodimeric complex comprising RUNX and tively and when deleted would drive tumorigenesis cofactor PEBP2b/CBFb 20 years ago, RUNX proteins (Bagchi and Mills, 2008). The 1p36 tumor suppressor are rapidly gaining prominence as major players in genes include -interacting zinc- cancer pathogenesis. In mammals, three RUNX family finger (RIZ)1, chromodomain helicase DNA-binding members have been identified. The RUNX genes are protein 5 (CHD5), secretory phospholipase PLA2G2A widely expressed, but show differential, tissue-specific (Modifier of Min-1) (Cormier et al., 1997), and TP73. expression. Although their strong homology indicates a Interestingly, RIZ1, also known as PRDM2,isa certain degree of redundancy as observed during mouse member of the nuclear histone/protein methyltransferase embryogenesis (Fukushima-Nakase et al., 2005) and superfamily that methylates histone 3 Lys 9 (H3K9) transcription regulation of various genes (for example residues, whereas CHD5 belongs to the SWI/SNF chromatin modifier family. As with all RUNX family Correspondence: Dr Y Ito, Institute of Molecular and Cell Biology, members, RUNX3 encodes a highly conserved Cancer Science Institute, National University of Singapore, 61 C-terminus motif VWRPY that was shown to bind Biopolis Drive, Proteos, Singapore 138673, Singapore. E-mail: [email protected] co-repressor proteins TLE1 and Sin3A, which in turn Received 4 November 2009; revised 12 February 2010; accepted 16 recruit histone deacetylase (Stifani et al., 1992; Aronson February 2010; published online 29 March 2010 et al., 1997; Imai et al., 1998; Levanon et al., 1998). RUNX3 and carcinogenesis of solid tumors LSH Chuang and Y Ito 2606 As part of this multiprotein complex, RUNX3 can is required for stem cell pluripotency and self-renewal. repress gene transcription through modulation of It is targeted to gene promoters of important develop- chromatin structure. It is tempting, therefore, to mental regulators, including RUNX3 (Lee et al., 2006), speculate that collaboration of RUNX3 with these in which it methylates histone H3 at the Lys27 (H3K27) other 1p36 chromatin modifiers suppresses transcription residue to repress transcription. Interestingly, EZH2 can that contributes to tumor development. direct DNA methylation by recruitment of DNMTs to target gene promoters, suggesting that it may be a significant causative factor of aberrant methylation in Epigenetic silencing of RUNX3 cancer (Vire et al., 2006). In addition, H3K27 methylation In tumor cells, mutation of one allele of a gene is has been shown to mark genes for de novo methylation in frequently accompanied by hypermethylation of the cancer (Schlesinger et al., 2007). A recent study found other allele, resulting in complete functional inactivation that frequent overexpression of EZH2 in aggressive of the gene (Jones and Baylin, 2002). Besides deletion of solid tumors is largely due to the loss of its regulator the RUNX3 gene, recent years have seen a surge in microRNA-101 (Varambally et al., 2008). More speci- reports of other modes of RUNX3 inactivation—be it fically, reintroduction of miR-101 or knockdown of through promoter hypermethylation or protein mislo- EZH2 expression was sufficient to reduce H3K27 calization—in solid tumor tissues. To date, hypermethy- trimethylation and increase RUNX3 mRNA levels. In lation of the CpG island at the RUNX3 promoter is the same vein, Fujii et al. (2008) found that EZH2 binds one of the most common aberrant methylation events to the RUNX3 promoter, resulting in upregulation of in cancer, suggesting that RUNX3 inactivation is a H3K27 methylation and concomitant downregulation significant risk factor for tumorigenesis. Diverse tumor of RUNX3 expression (Fujii et al., 2008). Recently, Lee tissues have been reported to exhibit RUNX3 hyper- et al. (2009) extended this finding by showing that methylation and inactivation. These include tissues hypoxia-induced upregulation of G9a histone methyl- and cell lines that originated from gastric (Li et al., transferase and HDAC1 expression also cause epige- 2002; Oshimo et al., 2004), bladder (Kim et al., 2005; netic RUNX3 silencing through H3K9 methylation and Wolff et al., 2008), colorectal (Ahlquist et al., 2008; decreased H3 acetylation at RUNX3 promoter. As Soong et al., 2009; Subramaniam et al., 2009), breast hypoxia is commonly associated with solid tumors of (Lau et al., 2006; Jiang et al., 2008), lung (Sato et al., 41mm3, this suggests that RUNX3 inactivation may be 2006), pancreatic (Wada et al., 2004), brain cancers associated with increased tumor size (Lee et al., 2009). (Mueller et al., 2007), and hepatocellular carcinoma Helicobacter pylori, a major causative factor of gastric (Kim et al., 2004). Notably, RUNX3 methylation status cancer, was also reported to alter RUNX3 methylation is one of the five markers used to classify colorectal status through enhanced production of nitric oxide in tumors associated with very high frequencies of CpG macrophages (Katayama et al., 2009). Moreover, the island methylation (CIMP), microsatellite instability, same group showed that lipopolysaccharide can also and BRAF mutation (Weisenberger et al., 2006). More- induce methylation changes, again through nitric oxide over, hypermethylation of RUNX3 is associated with a production, suggesting that inflammation may be 100-fold increase risk of developing bladder cancer another cause of RUNX3 hypermethylation in gastric (Kim et al., 2005). RUNX3 methylation is not only cancer. Estrogen exposure is an important risk factor in acquired early during tumorigenesis, but also increases breast cancer pathogenesis. It induces neoplastic trans- with age: RUNX3 is thus an attractive candidate for formation in epithelial cells that were derived from cancer detection and prognosis (Wolff et al., 2008). mammospheres isolated from healthy breast tissues. Conversely, RUNX3 overexpression correlates with Interestingly, exposure of these mammosphere-derived reduced invasive potential of breast cancer cells cells to estrogen was sufficient to induce hypermethyla- (Chen et al., 2007). Similarly, RUNX3 expression tion of RUNX3 (Cheng et al., 2008). All in all, these in the breast stroma is associated with reduced rate are clear indications that contextual cues from growth of recurrence, and thus, a good clinical outcome in factors and environmental stimuli give rise to aberrant breast cancer (Finak et al., 2008). Furthermore, the epigenetic silencing of RUNX3 in cancer cells (see presence of nuclear RUNX3 protein in esophageal Figure 1). squamous cell carcinomas correlates with increased sensitivity to radiotherapy, whereas RUNX3 inactiva- tion is linked to radioresistance and poor prognosis Cytoplasmic sequestration of RUNX3 (Sakakura et al., 2007). Not only is epigenetic silencing of RUNX3 a frequent occurrence in cancer, cytoplasmic sequestration of Putative causes of RUNX3 hypermethylation RUNX3 was reported in a significant proportion of Although it is unclear what causes aberrant methylation cancer cases—about 15% of colorectal cancer (Ito et al., of RUNX3, RUNX3 was recently reported to be silenced 2008), 80% of breast cancer (Lau et al., 2006), 38% of in gastric, breast, prostate, colon, and pancreatic cancer gastric cancer (Ito et al., 2005), and oral squamous cell cell lines by another epigenetic mechanism—enhancer of carcinomas (Gao et al., 2009). Moreover, RUNX3 Zeste Homologue 2 (EZH2)—mediated histone methy- cytoplasmic mislocalization is associated with advanced lation (Fujii et al., 2008). The EZH2 oncogene is a colorectal tumor stages, whereas nuclear RUNX3 component of Polycomb-repressor complex 2, which expression was correlated with better patient prognosis

Oncogene RUNX3 and carcinogenesis of solid tumors LSH Chuang and Y Ito 2607 DNA Hypermethylation However, it should be noted that RUNX3 cytoplasmic (Cheng et al., 2008) localization is not limited to cancer cells as a large proportion of chief cells from normal stomach Estrogen (Ito et al., 2005) and colon epithelia (Ito et al., 2008) show cytoplasmic localization of RUNX3. Thus, X CH3 CH3 elucidating this intrinsic ability of RUNX3 to shuttle between nuclear and cytoplasm compartments may offer RUNX3 insights as to how nuclear translocation of RUNX3 may be induced for cancer therapy. DNA Hypermethylation (Katayama et al., 2009) Helicobacter infection Nitric oxide Liposaccharide Mechanisms underlying cytoplasmic translocation Inflammation of RUNX3 X CH3 CH3 Interestingly, ectopic expression of a C-terminal trunca- RUNX3 tion of RUNX3 in SNU16 gastric cancer cells resulted in both nuclear exclusion of the exogenous RUNX3 Histone H3K9 methylation and Histone H3 deacetylation (1–187aa) deletion mutant and the endogenous wildtype (Lee et al., 2009) RUNX3 (Ito et al., 2005), suggesting that the C-terminal Hypoxia CH3 domain is necessary for nuclear localization of RUNX3. G9a HDAC1 Nuclear exclusion of RUNX proteins can be accom- H3K9 H3 Ac X modated by decreasing microtubule-dependent nuclear import, reducing nuclear retention by loss of association RUNX3 with chromatin-related macromolecular complexes or inhibiting nuclear export (Pockwinse et al., 2006; Pande Histone H3K27 methylation et al., 2009). Moreover, activation of TGF-b pathway (Fujii et al., 2008; Varambally et al., 2008) triggered nuclear translocation of endogenous RUNX3 in SNU16 cells and this subsequently led to growth EEZH2 CH3 inhibition (Ito et al., 2005). With the TGF-b pathway MicroRNA-101 EZH2 EZH2 frequently impaired in cancer tissues, this would mean H3K27 X frequent cytoplasmic sequestration and, thus, functional inactivation of RUNX3. It is not known how TGF-b RUNX3 elicits nuclear translocation of RUNX3. However, because SMAD and RUNX3 proteins interact (Hanai Figure 1 Mechanisms of epigenetic silencing of RUNX3. (a) Estrogen induces hypermethylation and silencing of RUNX3 et al., 1999), it is possible that perturbation of Smad in mammosphere-derived cells. (b) H. pylori infection may induce nuclear import/export may promote cytoplasmic accu- nitric oxide production, liposaccharide production, and inflamma- mulation (Inman et al., 2002). Alternatively, a clue can tion, leading to hypermethylation and silencing of RUNX3. be gleaned from the report that TGF-b stimulates (c) Hypoxia induces recruitment of G9a histone methyltransferase and HDAC1 to RUNX3, which in turn leads to histone 3 Lys 9 acetylation of RUNX3 protein by p300 and that this methylation and H3 deacetylation. (d) Overexpression of EZH2, inhibits RUNX3 ubiquitination by Smurf1, leading to which maybe the result of downregulation of microRNA-101, can the stabilization of RUNX3 protein (Jin et al., 2004). cause H3K27 methylation at RUNX3 promoter and subsequent It remains to be seen whether acetylated RUNX3 is transcription repression. more efficiently retained in the nucleus. Aside from acetylation, RUNX3 protein can be phosphorylated by the putative oncoprotein, Pim-1 serine/threonine kinase (Soong et al., 2009). Likewise, enforced cytoplasmic (Aho et al., 2006), and this induces nuclear exclusion of localization of RUNX3 is associated with increased RUNX3 (Kim et al., 2008). Furthermore, Jun-activation tumorigenic potential of gastric epithelial cells on domain-binding protein 1 (Jab1/CSN5) promotes nucle- engraftment onto nude mice (Ito et al., 2005). Collec- ar export and degradation of RUNX3 through CSN tively, these findings indicate that nuclear RUNX3 and (component of the COP9 signalosome)-associated ki- its downstream transcriptional targets are crucial for nase activities (Kim et al., 2009). Curiously, hypoxic tumor suppression. Sequestration of tumor suppressor culture conditions are also associated with cytoplasmic proteins in subcellular locations where they cannot retention of RUNX3 in a G9a and HDAC1-dependent exercise their function is a frequent occurrence in cancer manner. Consistent with this observation, overexpres- cells. Examples include well-established tumor suppres- sion of G9a and HDAC1 prevented nuclear transloca- sors such as p27KIP, BRCA1,andpRb1. As no mutation tion of RUNX3 (Lee et al., 2009). The underlying has been found in the coding sequence of the mis- mechanism remains unknown. More recently, onco- localized RUNX3, the reasons for cytoplasmic genes MDM2 E3 ubiquitin ligase and Src kinase were compartmentalization of RUNX3 is likely to be post- shown to facilitate cytoplasmic localization of RUNX3 translational modification or intracellular environment. (Chi et al., 2009; Goh et al., 2010). MDM2 mediates

Oncogene RUNX3 and carcinogenesis of solid tumors LSH Chuang and Y Ito 2608 Table 1 The different modes of RUNX3 inactivation in various tumor types Modes of RUNX3 Tumor types References inactivation

Hemizygous deletion Gastric, bile duct, pancreatic, lung cancers Li et al. (2002); Wada et al. (2004); Yanada et al. (2005) Mutations Gastric, bladder cancers Li et al. (2002); Kim et al. (2005) Hypermethylation Gastric, bladder, colorectal, breast, lung, hepatocel- Li et al. (2002); Kim et al. (2004); Oshimo et al. (2004); lular carcinoma, pancreatic, brain, prostate cancers Wada et al. (2004); Kim et al. (2005); Lau et al. (2006); Sato et al. (2006); Weisenberger et al. (2006); Mueller et al. (2007); Ahlquist et al. (2008); Cheng et al. (2008); Jiang et al. (2008); Wolff et al. (2008); Katayama et al. (2009); Soong et al. (2009); Subramaniam et al. (2009) Histone modifications Gastric, breast, prostate, colon and pancreatic cancer Fujii et al. (2008); Lee et al. (2009) Cytoplasmic mislocalization Colorectal, breast, gastric, oral squamous cell Ito et al. (2005); Lau et al. (2006); Ito et al. (2008); Gao carcinomas et al. (2009); Lee et al. (2009); Soong et al. (2009)

ubiquitination of RUNX3, resulting in its cytoplasmic and thus chronic myeloid leukemia (Miething translocation and degradation (Chi et al., 2009). Src et al., 2007). Runx3 was also identified as a cancer- kinase phosphorylates tyrosine residues in RUNX3, and associated gene by Sleeping Beauty transposable overexpression of activated Src correlates with RUNX3 elements—transposon integration at the 50 position of cytoplasmic localization (Goh et al., 2010). Collectively, the RUNX3 gene was associated with T-cell lymphoma these data indicate that aberrant signals from oncogenic (Dupuy et al., 2005) and prostate cancer (Rahrmann pathways may mediate post-translational modification et al., 2009). The non-viral Sleeping Beauty insertional and subsequent cytoplasmic mislocalization of RUNX3 mutagen thus offers a marked advantage over the in cancer cells. retroviral system, which is limited to identifying cancer genes in hematopoietic tumors: Sleeping Beauty screens can be adapted for epithelial cancers. This raises several interesting questions. How does this relate to other Insertional mutagenesis screen implicates RUNX3 cancers of epithelial origins in which RUNX3 functions in tumor development as tumor suppressor? Furthermore, as progenitor cancer stem cells are implicated in cancer persistence, does Several independent research groups have uncovered ectopic expression of RUNX3 influence the response of strong links between RUNX3 and cancer potentiation cancer stem cells to drug therapy? through insertional mutagenesis-based genetic screens. Altogether, these findings suggest that while RUNX3 It should be emphasized, however, that these screens inactivation is a major determinant of cancer patho- preferentially identify oncogenes and that among the genesis (see Table 1) and clinical outcome in many Runx family, Runx2 is most commonly reported cancer types, in certain tumor types, overexpression of (Cameron et al., 2003). An early indication of Runx3’s RUNX3 is oncogenic. In other words, RUNX3 involvement came from a study involving Moloney mediates context-dependent tumor suppression. murine leukemia virus infection of CD2- trans- genic mice (Stewart et al., 2002). The CD2-MYC mice selectively express the c-MYC oncogene in T cells. Although uninfected mice develop tumors infrequently, Molecular mechanisms that underlie RUNX3’s anti-tumor murine leukemia virus infection at birth accelerates properties tumor onset: all infected mice succumb to thymic lymphomas by 80 days (Stewart et al., 1996). After Runx3 knockout mice are prone to gastrointestinal cancer screening the viral-induced thymic lymphomas, a Although these different experimental approaches led to proviral integration site was uncovered 30 kb upstream the same conclusion—that of RUNX3’s involvement in of the Runx3 distal P1 promoter. This resulted in a cancer—they do not directly prove that RUNX3 is a significant increase in the expression of Runx3 P1 tumor suppressor in epithelial cells. In 2002, this concept (distal) isoforms in the pertinent lymphoma. As the was rigorously tested using Runx3 knockout mice only Runx protein expressed in this lymphoma, Runx3 (Li et al., 2002). Runx3À/À mice exhibited hyperplasia thus seems to synergize with MYC oncogene to promote of the gastric mucosa. The increased proliferation rate in lymphomagenesis (Stewart et al., 2002). Miething et al. the gastric epithelia of Runx3À/À mice was attributed to (2007) further explored RUNX3’s cancer function in suppression of apoptosis and reduced sensitivity to leukemia when they incorporated a drug used for the growth inhibitory effects of TGF-b1. RUNX3 treatment of Bcr-Abl-expressing leukemia, imatinib, in deficiency, thus, corresponded with compromised their retroviral mutagenesis screen—they showed that TGF-b signaling. Earlier shown to bind directly insertion of a retroviral element in the vicinity of the to TGF-b signaling effector SMAD3 for synergistic RUNX3 promoter results in elevated RUNX3, which in induction of Immunoglobulin germline C a promoter turn led to increased resistance to imatinib-induced (Shi and Stavnezer, 1998; Hanai et al., 1999) during

Oncogene RUNX3 and carcinogenesis of solid tumors LSH Chuang and Y Ito 2609 Immunoglogulin A class switching, RUNX3 is now also It is commonly accepted that progression from known to have a function in TGF-b-induced growth adenoma to carcinoma would require further inhibition. After stimulation by TGF-b cytokines, inappropriate inputs from other signaling networks. In RUNX3 cooperates with SMAD proteins to directly colorectal cancer, frequent mutations in various com- upregulate transcription of the proapoptotic gene Bim1 ponents of the TGF-b pathway impair the ability of and increase expression of the cyclin-dependent kinase cancer cells to undergo TGF-b-induced apoptosis. The inhibitor p21WAF1 in gastric cancer cells (Chi et al., 2005; ability of RUNX3 to regulate both Wnt and TGF-b Yano et al., 2006). Moreover, the ability of RUNX3 signaling pathways suggests that RUNX3 may orches- to target SMADs to distinct nuclear foci on stimulation trate tumor suppression in a three-way crosstalk. by TGF-b further strengthens the case for RUNX3 The isolation of the R122C mutation in RUNX3 from as a regulator of the TGF-b signaling pathway (Zaidi a gastric cancer patient was an important milestone— et al., 2002). this mutation confirms that dysfunctional RUNX3 can The detection of Runx3 in the epithelial cells of small indeed contribute to human tumor formation. The and large intestines in mice hinted that RUNX3 may also relevance of this mutation is underscored by the function as a tumor suppressor in the intestinal epithelium. discovery of a mutation (R169Q) at the equivalent Indeed, the intestinal epithelia of Runx3À/À mice exhibit position in RUNX2 from a patient with the skeletal hyperplasia that is associated with robust proliferation disorder cleidocranial dysplasia (Zhou et al., 1999). (Ito et al., 2008). Further examination revealed that this Notably, tumorigenicity of MKN28 gastric cancer- phenomenon is the likely consequence of enhanced Wnt derived cell line is strongly inhibited by exogenous signaling activity and transcriptional upregulation of expression of RUNX3. Exogenous expression of R122C Wnt target genes such as c-Myc and cyclinD1.RUNX3 not only failed to inhibit tumor growth in nude mice but was further found to form a ternary complex with key instead, induced larger tumors than the parental cell Wnt effectors TCF4 and b-catenin with the resulting line, suggesting that this single amino-acid substitution complex showing reduced ability to bind DNA and in the DNA-binding Runt domain was sufficient to activate transcription of Wnt target genes. RUNX3, confer an oncogenic function on Runx3 (Li et al., 2002). therefore, antagonizes aberrant Wnt signaling, which, This seems to stem from multiple defects in its protein substantial evidence have shown, provides an oncogenic properties. The R122C mutant has reduced affinity for impetus to sustain colorectal cancer growth. DNA with the consensus RUNX-binding site; it is also The Adenomatous Polyposis Coli (APC) gene is a less efficient in binding SMADs (Chi et al., 2005). The well-known gatekeeper of colorectal tumorigenesis. combination of these defects results in impaired To compare RUNX3’s tumor suppressive capabilities participation in TGF-b signaling, as seen by reduced against that of APC, Runx3 þ /À mice were bred in activation of p21WAF1 promoter during TGF-b stimula- parallel with ApcMin/ þ mice on the same BALB/c tion (Chi et al., 2005). More recently, we found that background. After a relatively long latency of about the R122C mutant is defective in binding to the 65 weeks of age, 54% of the Runx3 þ /À mice develop TCF4 (Ito et al., submitted), sug- small intestinal adenomas, a frequency comparable with gesting that this mutation enhances tumor growth that of ApcMin/ þ mice (64%) (Ito et al., 2008). Tumors because it impinges on the tumor suppressive fun- isolated from both mice displayed similar phenotypes— ctions of RUNX3 in both Wnt and TGF-b signaling upregulation of cyclin D1 and c-Myc, downregulation pathways during various stages of tumor progression. of p21WAF1—indicating increased activity of TCF4-b- Besides gastric cancer, mutations affecting the Runt catenin complex. Interestingly, unlike ApcMin/ þ mice domain of RUNX3 as well as RUNX3 polymorphisms in which biallelic loss of Apc occurs, the Apc locus have also been isolated from bladder cancer tissues was completely normal in Runx3 þ /À mice which (Kim et al., 2005; Zhang et al., 2008), suggesting exhibited no nuclear accumulation of oncogenic that such mutations or genetic variations may drive b-catenin. The fact that RUNX3 deficiency promotes tumorigenesis. adenoma formation independent of APC inactivation It is noteworthy that both Wnt and TGF-b signaling (Ito et al., 2008) indicates that RUNX3 is a gatekeeper pathways are also involved in the induction of epithelial- in its own right. mesenchymal transition (EMT) during embryonic de- Curiously, unlike Runx3 þ /À mice, Runx3À/À mice did velopment (Thiery and Sleeman, 2006). EMT is a not develop epithelial tumors. Although this is an complex developmental process in which epithelial cells unusual characteristic for a tumor suppressor gene, attain the migratory properties of mesenchymal cells there is a precedent: working with tumor suppressor through the loss of cell–cell or cell–matrix adhesion. Pten knockout mice, Pandolfi and co-workers observed Recently, the induction of EMT has been increasingly that although the loss of one allele of Pten is associated linked to cancer invasion and metastasis. Transcription with accelerated growth, the loss of both Pten alleles repressors Snail and Slug are believed to have important resulted instead in reduced proliferation, which stems functions in initiating the pathological EMT process. from cellular senescence (Chen et al., 2005). Pandolfi On oncogenic stimulation, they directly repress tran- suggested that biallelic inactivation of the Pten tumor scription of adherens-junction protein E-cadherin (Batlle suppressor elicits a strong oncogenic stimulus that sets et al., 2000; Cano et al., 2000) and tight-junction protein off a -mediated fail-safe mechanism to induce claudin-1 (Martinez-Estrada et al., 2006), facilitating senescence. loss of cell adhesion. Another inducer of EMT during

Oncogene RUNX3 and carcinogenesis of solid tumors LSH Chuang and Y Ito 2610 malignant progression is TGF-b signaling, which results epithelium (Brenner et al., 2004). We have earlier found in the dissolution of tight junctions. This ability of TGF-b that RUNX3 mRNA transcripts in GIT are relatively to induce EMT presents an interesting paradox that is low compared with peripheral blood (Ito et al., 2009). opposite of its tumor suppressive function. We found Moreover, RUNX3 was recently shown to interact with recently that claudin-1 is a direct and positive target of MDM2, which mediates ubiquitinylation and proteoly- Runx3 and that Runx3À/À-derived gastric epithelial tic degradation of RUNX3 (Chi et al., 2009). MDM2 is cells have reduced levels of claudin-1 and increased one of the E3 ligases in the ubiquitin-proteasome tumorigenic potential when compared with the Runx3 þ / system and is known to ubiquitinate and degrade p53 þ cells. This is partly because Runx3 cooperates with tumor suppressor. Owing to this, the amount of p53 is TGF-b effectors Smads to directly upregulate claudin-1 known to be quite low in normal cells (Ashcroft and transcription (Chang et al., 2010). As the expression of Vousden, 1999). It is thus likely that the already low claudin-1 inhibits proliferation in gastric cancer cells, expression level of Runx3 in GIT epithelium is further these findings present another mechanism underlying reduced because of MDM2-mediated degradation. Fail- RUNX3’s tumor suppressive effects and invite specula- ure to detect Runx3 expression in GIT, as reported by tion on the function of RUNX3 during later stages of Groner’s group, may stem from the low amounts of cancer. Whether RUNX3 affects the abilities of TGF-b Runx3 expressed in GIT epithelium and relative and Wnt to induce EMT remain unknown. sensitivity of various methods used for detection. Another way of explaining the oncogenicity of EMT The concept that inflammation promotes tumor is the ability of epithelial cells, but not mesenchymal development has steadily gained prominence in the cells, to undergo anoikis, an apoptotic process induced recent years (Coussens and Werb, 2002). For example, by cell detachment from the extracellular matrix. overexpression of a pro-inflammatory cytokine inter- Anoikis is, therefore, an important mechanism to leukin 1b in the gastric epithelium was sufficient to prevent metastasis. EMT processes promote anoikis- induce inflammation and hyperplasia (Tu et al., 2008). resistance, thus permitting tumorigenesis. The ability of Consistent with this notion, Groner’s group attributed RUNX3 to repress transcription of neutrophin the GIT hyperplasia exhibited by their Runx3 knockout TRKB during neuronal fate determination (Inoue et al., mice to inflammatory bowel disease caused by Runx3 2007) has important implications for RUNX3’s function deficiency in leukocytes in the GIT (Brenner et al., 2004; in EMT, considering that TRKB is a potent oncogene Levanon and Groner, 2009), as they could not detect that suppresses anoikis (Douma et al., 2004; Geiger and Runx3 expression in the GIT epithelium of wildtype Peeper, 2007). TRKB is highly expressed in neuroblastoma mice. In contrast, however, the knockout mice generated as well as aggressive tumors of pancreatic, prostate, and by Ito et al. (2008) exhibited minimal inflammation and lymphoid origins. Ectopic expression of TRKB was as such was unlikely to cause hyperplasia in the GIT sufficient to confer anoikis resistance and convert non- epithelium. Moreover, when bone marrow cells of malignant cells into highly tumorigenic cells. Induction Runx3À/À mice were transplanted into irradiated wild- of EMT is likely to coincide with reduction of RUNX3 type mice, these mice with Runx3À/À leukocytes and expression, followed by subsequent downregulation of wildtype epithelial cells neither exhibit significant infla- claudin-1, upregulation of TRKB, and suppression of mmation nor develop hyperplasia, further confirming anoikis. It is probable that RUNX3 represses TRKB that loss of Runx3 in epithelial cells led to acquisition of transcription to promote anoikis and suppress meta- tumorigenic potential in GIT (Ito et al., 2008). The static spread of cancer cells. varying degrees of inflammation induced in mice from Closely linked to EMT and the theme of progenitor the two laboratories may be caused by variations in the cell renewal is another exciting, but less well-defined conditions of the mouse facilities. concept—the function of cancer stem cells. EMT Although the studies by Ito et al. (2008) suggest that induction of transformed mammary epithelial cells leads hyperplasia of GIT in Runx3À/À mice is epithelial cell to enrichment of cancer stem cells (Mani et al., 2008). autonomous, the function of inflammation cannot be Together with the report that increased RUNX3 ignored. Runx3 is necessary for proper T-cell develop- expression is associated with cancer persistence in ment and it is possible that both Runx3À/À leukocytes CML (Miething et al., 2007), it may be that RUNX3 and epithelial cells contribute toward malignancy. is involved in cancer stem cell growth through modula- As the opposing viewpoints have already been tion of EMT processes. Moreover, these findings point discussed elsewhere, the reader is advised to refer to to the possibility that RUNX3 not only functions as a the respective publications (Bae and Ito, 2003; Levanon tumor suppressor during early tumor development, but et al., 2003; Ito et al., 2009; Levanon and Groner, 2009). may affect late stages of cancer progression. We anticipate that this controversy will soon be resolved with further in-depth studies.

Controversy on Runx3 expression in gastrointestinal tract epithelia RUNX3 has putative functions in other cancer signaling pathways It should be noted, however, that Groner and co- workers reported conflicting data: they could not detect The interaction of RUNX3 and putative oncoprotein Runx3 in the mouse gastrointestinal tract (GIT) yes-associated protein (YAP) through their PY

Oncogene RUNX3 and carcinogenesis of solid tumors LSH Chuang and Y Ito 2611 motif and WW domain, respectively, offers another p14ARF-MDM2 surveillance pathway in response to perspective (Yagi et al., 1999). RUNX3’s recruitment of aberrant Ras oncogene activation (Chi et al., 2009). YAP transcription co-activator results in enhanced target promoter activation. More recently, YAP was found to be a component of the Hippo pathway, which RUNX3 as a mediator of oncogene-induced senescence regulates organ size through modulation of proliferative and apoptotic activities. YAP expression is frequently Oncogene-induced senescence is a well-known charac- increased in cancer, and inactivation of YAP by the teristic of early preneoplastic lesions. Triggered by onco- Hippo pathway leads to reduced cell contact inhibition genes such as RAS through reactive oxygen species- (Zhao et al., 2007). As mutation of the WW domain in induced DNA damage and DNA hyperreplication YAP impairs its ability to promote cell proliferation and (Di Micco et al., 2006), oncogene-induced senescence oncogenic transformation (Zhao et al., 2009), it remains is an important mode of tumor suppression, which to be seen whether RUNX3 affects YAP tumor promot- blocks aberrant proliferative signals. Mounting evidence ing activities through interaction with its WW domain. have suggested cooperation between RUNX proteins The fact that RUNX3 physically interacts with and RAS pathways during tumorigenesis: concurrent Forkhead box O3a (FOXO3a) tumor suppressor protein mutations in RUNX1 and RAS pathways are extremely at the proapoptotic gene Bim promoter to promote frequent in RUNX1-related leukemia cases; Runx1 transcription and apoptosis (Yamamura et al., 2006) deficiency suppresses N-RAS-induced senescence and prompts speculation of functions for RUNX3 in other apoptosis, most likely through upregulation of Bmi-1 Foxo3A-related activities, and these include regulation and Bcl-2 and consequent suppression of p16Ink4a and of intestinal inflammation (Snoeks et al., 2009), AKT p19Arf expression (Motoda et al., 2007); RUNX3, similar oncogenic signaling pathway, response to oxidative to other RUNX family members, functions as a stress (Brunet et al.,2004),aswellasactivationofATM mediator of oncogene-induced senescence: it induces a during DNA damage (Tsai et al., 2008). Interestingly, p14ARF-p53-dependent growth arrest in primary mouse FOXOs protects hematopoietic stem cells from oxida- embryonic fibroblasts that is reminiscent of senescence tive stress and is thus necessary for long-term main- (Kilbey et al., 2007, 2008). Future studies may reveal the tenance of hematopoietic stem cells in the progenitor degree to which RUNX family members are biologically compartment (Tothova and Gilliland, 2007; Tothova linked to senescence. et al., 2007). Taken together with the recent report that resistance to reactive oxygen species is a characteristic of breast cancer stem cells (Diehn et al., 2009), it is worthwhile to investigate whether RUNX3 affects Inter-relationship of RUNX3 with other RUNX family FOXO activity during cancer stem cell formation. members

The strong conservation in amino-acid sequences of all RUNX family members suggests shared properties— RUNX3 in DNA repair they recognize the same DNA sequence—and some functional redundancy. Their differential expression It is well accepted that DNA damage promotes cancer profiles in various tissues, however, indicate that RUNX progression through genetic alterations of components proteins have distinct biological functions, depending on in cell signaling pathways, cell cycle regulation, repair, cell context. Although RUNX proteins do co-exist in and apoptotic processes. Certainly, dysfunctional some cells, inverse correlation of RUNX3 and RUNX1 repair mechanisms are associated with high cancer risks. expression levels were observed in lymphoblastoid cell Recent reports have linked RUNX3 with various DNA lines (Levanon et al., 1994; Spender et al., 2005): repair machineries. Mutations in components of DNA RUNX3 protein represses the RUNX1 P1 promoter mismatch repair are extremely frequent in colorectal through its C-terminal VWRPY amino-acid motif, cancer cells. In particular, deficiencies in MLH3 and likely in collaboration with TLE co-repressors (Spender PMS2 accelerate tumor progression. Against this et al., 2005; Brady et al., 2009). Moreover, this co- mismatch deficient background, overexpression of ordination of RUNX1 and RUNX3 levels is related to Transducin enhancer of Split (Tle) family gene, Tle6- cell proliferation, in which RUNX3-mediated repression like, further enhances tumor progression. The direct of RUNX1 promotes Epstein–Barr virus-induced interaction of RUNX3 with Tle6 results in attenuation immortalization of B cells (Brady and Farrell, 2009; of RUNX3 transactivation abilities (Chen et al., 2008). Brady et al., 2009). It would seem that RUNX3 and Aside from mismatch repair, Runx3 was found to RUNX1 serves contradictory functions: growth interact with DNA repair protein Ku70 (Tanaka et al., promoter and inhibitor, respectively. But how does this 2007), a component of the major repair mechanism for relate to solid tumors? Inverse correlation between double-strand DNA breaks, the non-homologous RUNX3 and RUNX2 expression was recently observed end-joining repair. The report detailing the interaction in some breast cancer cell lines in which cells with low of RUNX3 and MDM2 at the Runt domain (see RUNX3 expression show increased RUNX2 expression above) provides another interesting viewpoint that (Lau et al., 2006). As all RUNX proteins are closely RUNX3, in addition to p53, is negatively regulated by associated with cancer development, this communication

Oncogene RUNX3 and carcinogenesis of solid tumors LSH Chuang and Y Ito 2612 between family members has important implications: cells, reminiscent of the RUNX3-induced resistance to dysfunctional RUNX3 would affect RUNX1 and imatinib-induced apoptosis in chronic myeloid leukemia RUNX2 activities on cell proliferation. (Miething et al., 2007; Tsunematsu et al., 2009). Thus, although RUNX3 is more commonly observed to suppress tumor formation, it has a dualistic function in cancer—whether it acts as tumor suppressor or Oncogenic potential of RUNX3 oncogene is cell context dependent. How RUNX3 switches between such opposite activities remain unclear Paradoxically, similar to other members of the RUNX and is likely to be of great clinical relevance. family, RUNX3 can also function as an oncogene when overexpressed (Cameron and Neil, 2004). Strong evi- dence from murine retroviral insertional work (see above) is now corroborated with findings from clinical Conclusion samples and cancer cell lines. Immunohistochemical studies showed intense nuclear RUNX3 staining patterns In this review, we have highlighted important findings in basal cell carcinoma tissues (Salto-Tellez et al., 2006). that shape our current understanding of RUNX3 as a Moreover, RUNX3 protein is overexpressed in epithe- tumor suppressor (see Figure 2). Multiple lines of lial ovarian cancer. However, immunohistochemical evidence such as frequent inactivation of RUNX3 in staining revealed a striking difference—the upregulated cancer, its ability to interact with important components RUNX3 in ovarian cancer is cytoplasmic (Nevadunsky of various signaling pathways, and regulate their et al., 2009). Reintroduction of RUNX3 in ovarian output—as seen by its enhancement of TGF-b-related cancer cells led to increased viability, whereas depletion growth inhibition and suppression of oncogenic Wnt of RUNX3 resulted in reduced proliferation (Neva- activity—indicate that RUNX3 is closely linked to dunsky et al., 2009). RUNX3 also has an oncogenic cancer pathogenesis and that aberrant activity of function in head and neck squamous cell carcinoma in RUNX3 is a major determinant of cancer outcome. which RUNX3 expression was correlated with enhanced Furthermore, the rapidly expanding list of proteins that cell proliferation and malignancy (Tsunematsu et al., interacts with RUNX3 hints at involvement in other 2009). Interestingly, ectopic RUNX3 expression activities such as DNA damage response and repair. In inhibited chemotherapeutic drug adriamycin-induced time, a more complete knowledge of RUNX3’s multiple apoptosis in head and neck squamous cell carcinoma properties and functions will offer important insights

Wnt TGFβ/BMP

TrkB blocks anoikis

APC

SMAD4 SMAD4 β-catenin CBFβ TCF4 RUNX3 Anoikis RUNX3 cell cycle Cldn1 SMAD4 c-myc SMAD4 cyclinD1 EMT ? RUNX3 p21 CBFβ TrkB Cldn1 Bim

apoptosis

Figure 2 The pleiotropic effects of RUNX3 during tumor suppression. RUNX3 inhibits the oncogenic Wnt signaling pathway by forming a complex with the TCF4-b-catenin complex and preventing it from binding to target promoters such as c-Myc and cyclinD1; RUNX3 cooperates with SMAD3/SMAD4 to activate TGF-b-dependent growth inhibition and apoptosis by induction of p21 and Bim, respectively; RUNX3 induces Claudin-1 (Cldn-1) expression. Lack of RUNX3 may induce EMT; RUNX3 represses TrkB expression; in the absence of RUNX3, TRKB is overexpressed, which may prevent anoikis.

Oncogene RUNX3 and carcinogenesis of solid tumors LSH Chuang and Y Ito 2613 into the complex nature of tumor formation and allow Acknowledgements exploitation of its effects during cancer therapy. This work was funded by the Ministry of Education (MOE) Conflict of interest and National Research Foundation (NRF), Singapore. We give special thanks to Dr Kazuyoshi Kohu for his contribution The authors declare no conflict of interest. of Figure 2.

References

Ahlquist T, Lind GE, Costa VL, Meling GI, Vatn M, Hoff GS et al. Cheng AS, Culhane AC, Chan MW, Venkataramu CR, Ehrich M, (2008). Gene methylation profiles of normal mucosa, and benign Nasir A et al. (2008). Epithelial progeny of estrogen-exposed breast and malignant colorectal tumors identify early onset markers. Mol progenitor cells display a cancer-like methylome. Cancer Res 68: Cancer 7: 94. 1786–1796. Aho TL, Sandholm J, Peltola KJ, Ito Y, Koskinen PJ. (2006). Pim-1 Chi XZ, Kim J, Lee YH, Lee JW, Lee KS, Wee H et al. (2009). Runt- kinase phosphorylates RUNX family transcription factors and related transcription factor RUNX3 is a target of MDM2-mediated enhances their activity. BMC Cell Biol 7: 21. ubiquitination. Cancer Res 69: 8111–8119. Aronson BD, Fisher AL, Blechman K, Caudy M, Gergen JP. (1997). Chi XZ, Yang JO, Lee KY, Ito K, Sakakura C, Li QL et al. (2005). Groucho-dependent and -independent repression activities of Runt RUNX3 suppresses gastric epithelial cell growth by inducing domain proteins. Mol Cell Biol 17: 5581–5587. p21(WAF1/Cip1) expressi. Mol Cell Biol 25: 8097–8107. Ashcroft M, Vousden KH. (1999). Regulation of p53 stability. Cormier RT, Hong KH, Halberg RB, Hawkins TL, Richardson P, Oncogene 18: 7637–7643. Mulherkar R et al. (1997). Secretory phospholipase Pla2g2a confers Bae SC, Ito Y. (2003). Comment on Levanon et al., ‘Runx3 knockouts resistance to intestinal tumorigenesis. Nat Genet 17: 88–91. and stomach cancer’, in EMBO reports (June 2003). EMBO Rep 4: Coussens LM, Werb Z. (2002). Inflammation and cancer. Nature 420: 538–539. 860–867. Bagchi A, Mills AA. (2008). The quest for the 1p36 tumor suppressor. Di Micco R, Fumagalli M, Cicalese A, Piccinin S, Gasparini P, Luise Cancer Res 68: 2551–2556. C et al. (2006). Oncogene-induced senescence is a DNA damage Batlle E, Sancho E, Franci C, Dominguez D, Monfar M, Baulida J response triggered by DNA hyper-replication. Nature 444: 638–642. et al. (2000). The transcription factor snail is a repressor of E- Diehn M, Cho RW, Lobo NA, Kalisky T, Dorie MJ, Kulp AN et al. cadherin in epithelial tumour cells. Nat Cell Biol 2: (2009). Association of reactive oxygen species levels and radio- 84–89. resistance in cancer stem cells. Nature 458: 780–783. Brady G, Farrell PJ. (2009). RUNX3-mediated repression of RUNX1 Douma S, Van Laar T, Zevenhoven J, Meuwissen R, Van Garderen E, in B cells. J Cell Physiol 221: 283–287. Peeper DS. (2004). Suppression of anoikis and induction Brady G, Whiteman HJ, Spender LC, Farrell PJ. (2009). Down- of metastasis by the neurotrophic receptor TrkB. Nature 430: regulation of RUNX1 by RUNX3 requires the RUNX3 VWRPY 1034–1039. sequence and is essential for Epstein-Barr virus-driven B-cell Dupuy AJ, Akagi K, Largaespada DA, Copeland NG, Jenkins NA. proliferation. J Virol 83: 6909–6916. (2005). Mammalian mutagenesis using a highly mobile somatic Brenner O, Levanon D, Negreanu V, Golubkov O, Fainaru O, Sleeping Beauty transposon system. Nature 436: 221–226. Woolf E et al. (2004). Loss of Runx3 function in leukocytes is Finak G, Bertos N, Pepin F, Sadekova S, Souleimanova M, Zhao H associated with spontaneously developed colitis and gastric mucosal et al. (2008). Stromal gene expression predicts clinical outcome in hyperplasia. Proc Natl Acad Sci USA 101: 16016–16021. breast cancer. Nat Med 14: 518–527. Brunet A, Sweeney LB, Sturgill JF, Chua KF, Greer PL, Lin Y et al. Fujii S, Ito K, Ito Y, Ochiai A. (2008). Enhancer of zeste homologue 2 (2004). Stress-dependent regulation of FOXO transcription factors (EZH2) down-regulates RUNX3 by increasing histone H3 methyla- by the SIRT1 deacetylase. Science 303: 2011–2015. tion. J Biol Chem 283: 17324–17332. Cameron ER, Blyth K, Hanlon L, Kilbey A, Mackay N, Stewart M Fukushima-Nakase Y, Naoe Y, Taniuchi I, Hosoi H, Sugimoto T, et al. (2003). The Runx genes as dominant oncogenes. Blood Cells Okuda T. (2005). Shared and distinct roles mediated through C- Mol Dis 30: 194–200. terminal subdomains of acute myeloid leukemia/Runt-related Cameron ER, Neil JC. (2004). The Runx genes: lineage-specific transcription factor molecules in murine development. Blood 105: oncogenes and tumor suppressors. Oncogene 23: 4308–4314. 4298–4307. Cano A, Perez-Moreno MA, Rodrigo I, Locascio A, Blanco MJ, del Gao F, Huang C, Lin M, Wang Z, Shen J, Zhang H et al. (2009). Barrio MG et al. (2000). The transcription factor snail controls Frequent inactivation of RUNX3 by promoter hypermethylation epithelial-mesenchymal transitions by repressing E-cadherin expres- and protein mislocalization in oral squamous cell carcinomas. sion. Nat Cell Biol 2: 76–83. J Cancer Res Clin Oncol 135: 739–747. Chang TL, Ito K, Ko TK, Liu Q, Salto-Tellez M, Yeoh KG et al. Geiger TR, Peeper DS. (2007). Critical role for TrkB kinase function in (2010). Claudin-1 has tumor suppressive activity and is a direct anoikis suppression, tumorigenesis, and metastasis. Cancer Res 67: target of RUNX3 in gastric epithelial cells. Gastroenterology 138: 6221–6229. 255–265 e1-3. Goh YM, Cinghu S, Hwee Hong ET, Lee YS, Kim JH, Zhang JW Chen PC, Kuraguchi M, Velasquez J, Wang Y, Yang K, Edwards R et al. (2010). Src kinase phosphorylates RUNX3 at tyrosine residues et al. (2008). Novel roles for MLH3 deficiency and TLE6-like and localizes the protein in the cytoplasm. J Biol Chem (e-pub ahead amplification in DNA mismatch repair-deficient gastrointestinal of print 25 January 2010; doi:10.1074/jbc.M109.071381). tumorigenesis and progression. PLoS Genet 4: e1000092. Hanai J, Chen LF, Kanno T, Ohtani-Fujita N, Kim WY, Guo WH Chen W, Salto-Tellez M, Palanisamy N, Ganesan K, Hou Q, Tan LK et al. (1999). Interaction and functional cooperation of PEBP2/CBF et al. (2007). Targets of genome copy number reduction in primary with Smads. Synergistic induction of the immunoglobulin germline breast cancers identified by integrative genomics. Genes Chromo- Calpha promoter. J Biol Chem 274: 31577–31582. somes Cancer 46: 288–301. Imai Y, Kurokawa M, Tanaka K, Friedman AD, Ogawa S, Mitani K Chen Z, Trotman LC, Shaffer D, Lin HK, Dotan ZA, Niki M et al. et al. (1998). TLE, the human homolog of groucho, interacts with (2005). Crucial role of p53-dependent cellular senescence in AML1 and acts as a repressor of AML1-induced transactivation. suppression of Pten-deficient tumorigenesis. Nature 436: 725–730. Biochem Biophys Res Commun 252: 582–589.

Oncogene RUNX3 and carcinogenesis of solid tumors LSH Chuang and Y Ito 2614 Inman GJ, Nicolas FJ, Hill CS. (2002). Nucleocytoplasmic shuttling of Li QL, Ito K, Sakakura C, Fukamachi H, Inoue K, Chi XZ et al. Smads 2, 3, and 4 permits sensing of TGF-beta receptor activity. (2002). Causal relationship between the loss of RUNX3 expression Mol Cell 10: 283–294. and gastric cancer. Cell 109: 113–124. Inoue K, Ito K, Osato M, Lee B, Bae SC, Ito Y. (2007). The Mani SA, Guo W, Liao MJ, Eaton EN, Ayyanan A, Zhou AY et al. transcription factor Runx3 represses the neurotrophin receptor (2008). The epithelial-mesenchymal transition generates cells with TrkB during lineage commitment of dorsal root ganglion neurons. properties of stem cells. Cell 133: 704–715. J Biol Chem 282: 24175–24184. Martinez-Estrada OM, Culleres A, Soriano FX, Peinado H, Bolos V, Ito K, Inoue KI, Bae SC, Ito Y. (2009). Runx3 expression in Martinez FO et al. (2006). The transcription factors Slug and Snail gastrointestinal tract epithelium: resolving the controversy. Onco- act as repressors of Claudin-1 expression in epithelial cells. Biochem gene 28: 1379–1384. J 394: 449–457. Ito K, Lim AC, Salto-Tellez M, Motoda L, Osato M, Chuang LS et al. Miething C, Grundler R, Mugler C, Brero S, Hoepfl J, Geigl J et al. (2008). RUNX3 attenuates beta-catenin/T cell factors in intestinal (2007). Retroviral insertional mutagenesis identifies RUNX genes tumorigenesis. Cancer Cell 14: 226–237. involved in chronic myeloid leukemia disease persistence under Ito K, Liu Q, Salto-Tellez M, Yano T, Tada K, Ida H et al. (2005). imatinib treatment. Proc Natl Acad Sci USA 104: 4594–4599. RUNX3, a novel tumor suppressor, is frequently inactivated in Motoda L, Osato M, Yamashita N, Jacob B, Chen LQ, Yanagida M gastric cancer by protein mislocalization. Cancer Res 65: 7743–7750. et al. (2007). Runx1 protects hematopoietic stem/progenitor cells Jiang Y, Tong D, Lou G, Zhang Y, Geng J. (2008). Expression of from oncogenic insult. Stem Cells 25: 2976–2986. RUNX3 gene, methylation status and clinicopathological signifi- Mueller W, Nutt CL, Ehrich M, Riemenschneider MJ, von Deimling cance in breast cancer and breast cancer cell lines. Pathobiology 75: A, van den Boom D et al. (2007). Downregulation of RUNX3 and 244–251. TES by hypermethylation in glioblastoma. Oncogene 26: 583–593. Jin YH, Jeon EJ, Li QL, Lee YH, Choi JK, Kim WJ et al. (2004). Nevadunsky NS, Barbieri JS, Kwong J, Merritt MA, Welch WR, Transforming growth factor-beta stimulates p300-dependent Berkowitz RS et al. (2009). RUNX3 protein is overexpressed in RUNX3 acetylation, which inhibits ubiquitination-mediated degra- human epithelial ovarian cancer. Gynecol Oncol 112: 325–330. dation. J Biol Chem 279: 29409–29417. Oshimo Y, Oue N, Mitani Y, Nakayama H, Kitadai Y, Yoshida K Jones PA, Baylin SB. (2002). The fundamental role of epigenetic events et al. (2004). Frequent loss of RUNX3 expression by promoter in cancer. Nat Rev Genet 3: 415–428. hypermethylation in gastric carcinoma. Pathobiology 71: 137–143. Katayama Y, Takahashi M, Kuwayama H. (2009). Helicobacter pylori Pande S, Ali SA, Dowdy C, Zaidi SK, Ito K, Ito Y et al. (2009). causes gene methylation and its loss of expression in gastric Subnuclear targeting of the Runx3 tumor suppressor and its epithelial cells, which is mediated by nitric oxide produced by epigenetic association with mitotic . J Cell Physiol macrophages. Biochem Biophys Res Commun 388: 496–500. 218: 473–479. Kilbey A, Blyth K, Wotton S, Terry A, Jenkins A, Bell M et al. (2007). Pockwinse SM, Rajgopal A, Young DW, Mujeeb KA, Nickerson J, Runx2 disruption promotes immortalization and confers resistance Javed A et al. (2006). Microtubule-dependent nuclear-cytoplasmic to oncogene-induced senescence in primary murine fibroblasts. shuttling of Runx2. J Cell Physiol 206: 354–362. Cancer Res 67: 11263–11271. Rahrmann EP, Collier LS, Knutson TP, Doyal ME, Kuslak SL, Green Kilbey A, Terry A, Cameron ER, Neil JC. (2008). Oncogene-induced LE et al. (2009). Identification of PDE4D as a proliferation senescence: an essential role for Runx. Cell Cycle 7: 2333–2340. promoting factor in prostate cancer using a Sleeping Beauty Kim HR, Oh BC, Choi JK, Bae SC. (2008). Pim-1 kinase transposon-based somatic mutagenesis screen. Cancer Res 69: phosphorylates and stabilizes RUNX3 and alters its subcellular 4388–4397. localization. J Cell Biochem 105: 1048–1058. Sakakura C, Miyagawa K, Fukuda KI, Nakashima S, Yoshikawa T, Kim JH, Choi JK, Cinghu S, Jang JW, Lee YS, Li YH et al. (2009). Kin S et al. (2007). Frequent silencing of RUNX3 in esophageal Jab1/CSN5 induces the cytoplasmic localization and degradation of squamous cell carcinomas is associated with radioresistance and RUNX3. J Cell Biochem 107: 557–565. poor prognosis. Oncogene 26: 5927–5938. Kim TY, Lee HJ, Hwang KS, Lee M, Kim JW, Bang YJ et al. (2004). Salto-Tellez M, Peh BK, Ito K, Tan SH, Chong PY, Han HC et al. Methylation of RUNX3 in various types of human cancers and (2006). RUNX3 protein is overexpressed in human basal cell premalignant stages of gastric carcinoma. Lab Invest 84: 479–484. carcinomas. Oncogene 25: 7646–7649. Kim WJ, Kim EJ, Jeong P, Quan C, Kim J, Li QL et al. (2005). Sato K, Tomizawa Y, Iijima H, Saito R, Ishizuka T, Nakajima T et al. RUNX3 inactivation by point mutations and aberrant DNA (2006). Epigenetic inactivation of the RUNX3 gene in lung cancer. methylation in bladder tumors. Cancer Res 65: 9347–9354. Oncol Rep 15: 129–135. Lau QC, Raja E, Salto-Tellez M, Liu Q, Ito K, Inoue M et al. (2006). Schlesinger Y, Straussman R, Keshet I, Farkash S, Hecht M, RUNX3 is frequently inactivated by dual mechanisms of protein Zimmerman J et al. (2007). Polycomb-mediated methylation on mislocalization and promoter hypermethylation in breast cancer. Lys27 of histone H3 pre-marks genes for de novo methylation in Cancer Res 66: 6512–6520. cancer. Nat Genet 39: 232–236. Lee SH, Kim J, Kim WH, Lee YM. (2009). Hypoxic silencing of tumor Shi MJ, Stavnezer J. (1998). CBF alpha3 (AML2) is induced by suppressor RUNX3 by histone modification in gastric cancer cells. TGF-beta1 to bind and activate the mouse germline Ig alpha Oncogene 28: 184–194. promoter. J Immunol 161: 6751–6760. Lee TI, Jenner RG, Boyer LA, Guenther MG, Levine SS, Kumar RM Snoeks L, Weber CR, Wasland K, Turner JR, Vainder C, Qi W et al. et al. (2006). Control of developmental regulators by Polycomb in (2009). Tumor suppressor FOXO3 participates in the regulation of human embryonic stem cells. Cell 125: 301–313. intestinal inflammation. Lab Invest 89: 1053–1062. Levanon D, Brenner O, Otto F, Groner Y. (2003). Runx3 knockouts Soong R, Shah N, Peh BK, Chong PY, Ng SS, Zeps N et al. (2009). and stomach cancer. EMBO Rep 4: 560–564. The expression of RUNX3 in colorectal cancer is associated with Levanon D, Goldstein RE, Bernstein Y, Tang H, Goldenberg D, disease stage and patient outcome. Br J Cancer 100: 676–679. Stifani S et al. (1998). Transcriptional repression by AML1 and Spender LC, Whiteman HJ, Karstegl CE, Farrell PJ. (2005). LEF-1 is mediated by the TLE/Groucho corepressors. Proc Natl Transcriptional cross-regulation of RUNX1 by RUNX3 in human Acad Sci USA 95: 11590–11595. B cells. Oncogene 24: 1873–1881. Levanon D, Groner Y. (2009). Runx3-deficient mouse strains circa Stewart M, MacKay N, Cameron ER, Neil JC. (2002). The common 2008: resemblance and dissimilarity. Blood Cells Mol Dis 43: 1–5. retroviral insertion locus Dsi1 maps 30 kilobases upstream of Levanon D, Negreanu V, Bernstein Y, Bar-Am I, Avivi L, Groner Y. the P1 promoter of the murine Runx3/Cbfa3/Aml2 gene. J Virol 76: (1994). AML1, AML2, and AML3, the human members of the runt 4364–4369. domain gene-family: cDNA structure, expression, and chromoso- Stewart M, Terry A, O’Hara M, Cameron E, Onions D, Neil JC. mal localization. Genomics 23: 425–432. (1996). til-1: a novel proviral insertion locus for Moloney murine

Oncogene RUNX3 and carcinogenesis of solid tumors LSH Chuang and Y Ito 2615 leukaemia virus in lymphomas of CD2-myc transgenic mice. J Gen underlies sporadic microsatellite instability and is tightly associated Virol 77(Part 3): 443–446. with BRAF mutation in colorectal cancer. Nat Genet 38: 787–793. Stifani S, Blaumueller CM, Redhead NJ, Hill RE, Artavanis-Tsakonas Wolff EM, Liang G, Cortez CC, Tsai YC, Castelao JE, Cortessis VK S. (1992). Human homologs of a Drosophila Enhancer of split et al. (2008). RUNX3 methylation reveals that bladder tumors gene product define a novel family of nuclear proteins. Nat Genet 2: are older in patients with a history of smoking. Cancer Res 68: 119–127. 6208–6214. Subramaniam MM, Chan JY, Soong R, Ito K, Yeoh KG, Wong R Wotton S, Terry A, Kilbey A, Jenkins A, Herzyk P, Cameron E et al. et al. (2009). RUNX3 inactivation in colorectal polyps arising (2008). Gene array analysis reveals a common Runx transcriptional through different pathways of colonic carcinogenesis. Am J programme controlling cell adhesion and survival. Oncogene 27: Gastroenterol 104: 426–436. 5856–5866. Tanaka Y, Imamura J, Kanai F, Ichimura T, Isobe T, Koike M et al. Yagi R, Chen LF, Shigesada K, Murakami Y, Ito Y. (1999). A WW (2007). Runx3 interacts with DNA repair protein Ku70. Exp Cell domain-containing yes-associated protein (YAP) is a novel tran- Res 313: 3251–3260. scriptional co-activator. EMBO J 18: 2551–2562. Thiery JP, Sleeman JP. (2006). Complex networks orchestrate Yamamura Y, Lee WL, Inoue K, Ida H, Ito Y. (2006). RUNX3 epithelial-mesenchymal transitions. Nat Rev Mol Cell Biol 7: cooperates with FoxO3a to induce apoptosis in gastric cancer cells. 131–142. J Biol Chem 281: 5267–5276. Tothova Z, Gilliland DG. (2007). FoxO transcription factors and stem Yanada M, Yaoi T, Shimada J, Sakakura C, Nishimura M, Ito K et al. cell homeostasis: insights from the hematopoietic system. Cell Stem (2005). Frequent hemizygous deletion at 1p36 and hypermethylation Cell 1: 140–152. downregulate RUNX3 expression in human lung cancer cell lines. Tothova Z, Kollipara R, Huntly BJ, Lee BH, Castrillon DH, Cullen Oncol Rep 14: 817–822. DE et al. (2007). FoxOs are critical mediators of hematopoietic stem Yano T, Ito K, Fukamachi H, Chi XZ, Wee HJ, Inoue K et al. (2006). cell resistance to physiologic oxidative stress. Cell 128: 325–339. The RUNX3 tumor suppressor upregulates Bim in gastric epithelial Tsai WB, Chung YM, Takahashi Y, Xu Z, Hu MC. (2008). Functional cells undergoing transforming growth factor beta-induced interaction between FOXO3a and ATM regulates DNA damage apoptosis. Mol Cell Biol 26: 4474–4488. response. Nat Cell Biol 10: 460–467. Zaidi SK, Sullivan AJ, van Wijnen AJ, Stein JL, Stein GS, Lian JB. Tsunematsu T, Kudo Y, Iizuka S, Ogawa I, Fujita T, Kurihara H et al. (2002). Integration of Runx and Smad regulatory signals at (2009). RUNX3 has an oncogenic role in head and neck cancer. transcriptionally active subnuclear sites. Proc Natl Acad Sci USA PLoS One 4: e5892. 99: 8048–8053. Tu S, Bhagat G, Cui G, Takaishi S, Kurt-Jones EA, Rickman B et al. Zhang Z, Wang S, Wang M, Tong N, Fu G. (2008). Genetic variants in (2008). Overexpression of interleukin-1beta induces gastric inflam- RUNX3 and risk of bladder cancer: a haplotype-based analysis. mation and cancer and mobilizes myeloid-derived suppressor cells in Carcinogenesis 29: 1973–1978. mice. Cancer Cell 14: 408–419. Zhao B, Kim J, Ye X, Lai ZC, Guan KL. (2009). Both TEAD-binding Varambally S, Cao Q, Mani RS, Shankar S, Wang X, Ateeq B et al. and WW domains are required for the growth stimulation and (2008). Genomic loss of microRNA-101 leads to overexpression of oncogenic transformation activity of yes-associated protein. Cancer histone methyltransferase EZH2 in cancer. Science 322: 1695–1699. Res 69: 1089–1098. Vire E, Brenner C, Deplus R, Blanchon L, Fraga M, Didelot C et al. Zhao B, Wei X, Li W, Udan RS, Yang Q, Kim J et al. (2007). (2006). The Polycomb group protein EZH2 directly controls DNA Inactivation of YAP oncoprotein by the Hippo pathway is involved methylation. Nature 439: 871–874. in cell contact inhibition and tissue growth control. Genes Dev 21: Wada M, Yazumi S, Takaishi S, Hasegawa K, Sawada M, Tanaka H 2747–2761. et al. (2004). Frequent loss of RUNX3 gene expression in human Zhou G, Chen Y, Zhou L, Thirunavukkarasu K, Hecht J, Chitayat D bile duct and pancreatic cancer cell lines. Oncogene 23: 2401–2407. et al. (1999). CBFA1 mutation analysis and functional correlation Weisenberger DJ, Siegmund KD, Campan M, Young J, Long TI, with phenotypic variability in cleidocranial dysplasia. Hum Mol Faasse MA et al. (2006). CpG island methylator phenotype Genet 8: 2311–2316.

Oncogene