Oncogene (2010) 29, 4671–4681 & 2010 Macmillan Publishers Limited All rights reserved 0950-9232/10 www.nature.com/onc ORIGINAL ARTICLE DEK oncoprotein regulates transcriptional modifiers and sustains tumor initiation activity in high-grade neuroendocrine carcinoma of the lung

T Shibata1,2, A Kokubu1, M Miyamoto1, F Hosoda1, M Gotoh2, K Tsuta3, H Asamura4, Y Matsuno5, T Kondo6, I Imoto7,8, J Inazawa7,8 and S Hirohashi1,2

1Cancer Genomics Project, National Cancer Center Research Institute, Chuo-ku, Tokyo, Japan; 2Pathology Division, National Cancer Center Research Institute, Chuo-ku, Tokyo, Japan; 3Clinical Laboratory Division, National Cancer Center Hospital, Chuo-ku, Tokyo, Japan; 4Division of Thoracic Surgery, National Cancer Center Hospital, Chuo-ku, Tokyo, Japan; 5Department of Surgical Pathology, Hokkaido University Hospital, Sapporo, Hokkaido, Japan; 6Proteome Bioinfomatics Project, National Cancer Center Research Institute, Chuo-ku, Tokyo, Japan; 7Department of Molecular Cytogenetics, Medical Research Institute and Graduate School of Biomedical Science, Tokyo Medical and Dental University, Tokyo, Japan and 8Center of Excellence Program for Frontier Research on Molecular Destruction and Reconstitution of Tooth and Bone, Tokyo Medical and Dental University, Tokyo, Japan

Lung cancer shows diverse histological subtypes. Large- Introduction cell neuroendocrine cell carcinoma and small-cell lung carcinoma show similar histological features and clinical Lung cancer is one of the most prevalent cancers behaviors, and can be classified as high-grade neuro- worldwide, and shows diverse histological subtypes, endocrine carcinoma (HGNEC) of the lung. Here we including adenocarcinoma, squamous cell carcinoma, elucidated the molecular classification of pulmonary large-cell carcinoma, large-cell neuroendocrine carcino- endocrine tumors by copy-number profiling. We compared ma (LCNEC) and small-cell lung carcinoma (SCLC; alterations of copy number with the clinical outcome of Travis and Brambilla, 2004). The latter two share HGNEC and identified a chromosomal gain of the DEK common histological features such as neuroendocrine oncogene locus (6p22.3) that was significantly associated differentiation and belong to a spectrum of endocrine with poor prognosis. We further confirmed that DEK neoplasms of the lung, which include benign (typical overexpression was associated with poor prognosis in a carcinoid; TC), intermediately malignant (atypical larger set of HGNEC. Downregulation of DEK by small carcinoid; AC) and highly malignant (LCNEC and hairpin RNA led to a marked reduction of in vitro colony SCLC) tumors (Travis et al., 1991; Travis and Brambil- formation, in vivo tumorigenicity and chemo-resistance, la, 2004). LCNEC and SCLC show similar clinical and was associated with loss of lung cancer stem cell behavior and can be classified as high-grade neuro- markers. Gene expression profiling revealed that DEK endocrine carcinoma (HGNEC; Asamura et al., 2006). downregulation was associated with altered expression of Genome-wide expression profiling has supported this transcriptional regulators, which specifically include classification, and led to the division of HGNEC into known targets of interchromosomal translocations in groups showing better and poor prognosis, showing no hematopoietic tumors, and knockdown of these epigenetic correlation with histological subtype (LCNEC or SCLC; modifiers affected colony formation activity. Our study Jones et al., 2004). However, details of the molecular showed that DEK overexpression, partly through an alterations in HGNEC, especially those associated with increase in its gene dose, mediates the activity of global malignant phenotypes, remain largely unknown. transcriptional regulators and is associated with tumor Recently, many studies have reported that a self- initiation activity and poor prognosis in HGNEC. renewing population of cancer cells, known as cancer Oncogene (2010) 29, 4671–4681; doi:10.1038/onc.2010.217; stem cell (CSC) or tumor-initiating cell, is responsible published online 14 June 2010 for histological heterogeneity, tumor recurrence, organ metastasis and drug resistance, which are hallmarks of Keywords: lung cancer; cancer stem cell; DEK; malignancy and related to poor prognosis (Visvader and neuroendocrine Lindeman, 2008). Interestingly, HGNEC is well known to frequently coexist with other histological subtypes such as adenocarcinoma or squamous cell carcinoma (so-called combined tumor; Travis and Brambilla, 2004), suggesting that this tumor contains a pluripotent stem cell-like population. Although several putative markers for lung cancer stem cells (such as CD133 and Correspondence: Dr T Shibata, Cancer Genomics Project, National ALDH1A) have been proposed (Peacock and Watkins, Cancer Center Research Institute, 5-1-1, Tsukiji, Chuo-ku, Tokyo 2008), the molecular mechanisms responsible for main- 104-0045, Japan, E-mail: [email protected] taining these stem cell features are largely unknown. Received 29 September 2009; revised 26 April 2010; accepted 2 May To further understand the molecular mechanisms 2010; published online 14 June 2010 related to the progression of HGNEC, we have DEK oncoprotein in pulmonary endocrine carcinoma T Shibata et al 4672 compared copy-number alterations with clinical out- Thus, the copy number-based classification elucidated a come of HGNEC and shown that chromosomal gain of carcinoid subgroup, which contains all carcinoid sam- the DEK gene locus is significantly associated with poor ples and exceptionally one LCNEC case (Carcinoid prognosis. DEK was originally identified as a fusion branch in Figure 1b). Unsupervised classification partner of t(6;9)(p23;q34) translocation in a subtype of revealed three subclasses (HGNEC-BR1-3) in the acute myeloid leukemia; von Lindern et al., 1992) and is HGNEC group, suggesting of the possibility of mole- overexpressed in several types of solid tumor such as cular subtypes in HGNEC, and SCLC was significantly liver cancer (Kondoh et al., 1999; Carro et al., 2006; segregated in HGNEC-BR1 (P ¼ 0.00246). We then Khodadoust et al.,2009). performed extensive molecular characterization of the lung endocrine tumors by the immunohistochemical analysis of candidate tumor suppressor (p53, PTEN and Results RB) and oncogene (SKP2, EGFR, MET and KIT) products in this cohort (Figure 1b). Aberrant expres- Molecular classification of endocrine tumors of the lung sions of these proteins were more frequent in HGNEC, To clarify the chromosomal alteration profile of the but no significant correlation was observed among endocrine tumors of the lung, we performed array-based HGNEC-branches and oncoprotein expressions (data comparative genomic hybridization (CGH) analysis of not shown). Notwithstanding the inability to character- the 49 primary tumors (11 cases of TC, 2 cases of AC, 8 ize this classification by known molecular markers, cases of SCLC and 28 cases of LCNEC), which include when clinical outcome was compared, patients in samples analyzed in our previous report (Peng et al., HGNEC-BR2 showed a significantly better outcome 2005). We then conducted unsupervised cluster analysis than the others (HGNEC-BR1/3; Figure 1c and of the CGH data to examine whether histological Supplementary Figure 1), suggesting of novel molecular subtypes have any association with genomic classifica- markers that can discriminate the two groups. Then we tion (Figure 1a). As shown in Figures 1a and b, searched for the genetic differences between HGNEC- chromosome alterations in TC and AC are characteristic BR1/3 (poor prognosis) and HGNEC-BR2 (good and clearly distinct from those observed in HGNEC prognosis) and found that copy number alterations at (LCNEC and SCLC; distinctive bacterial artificial four loci (12q13 EPS8 (adjusted P ¼ 0.0033), 6p22.3 chromosome (BAC) clones between carcinoid and DEK (P ¼ 0.017), 2p24 NBAS (P ¼ 0.022) and 2q31.3 HGNEC are listed in the Supplementary Table 1). ITGA4 (P ¼ 0.046)) differed significantly between the

Figure 1 Molecular classification of pulmonary endocrine tumors. (a) Hierarchical cluster analysis of pulmonary endocrine tumors based on the changes in copy number at 800 loci. Red indicates copy number gain/amplification and green indicates copy number loss. Color bars at the bottom show subgroups. (b) Association between immunohistological features and genetic classification in pulmonary endocrine tumors. Each column indicates a single primary tumor. In the histology row, blue, yellow, red and green boxes indicate TC, AC, LCNEC and SCLC, respectively. In the p53 row, solid and gray boxes indicate strong and moderate p53 protein expression, respectively, which are suggestive of mutant p53. In the PTEN and RB rows, solid box indicates loss of tumor-suppressor protein expression. In other rows, solid and gray boxes indicate strong and moderate oncoprotein expression, respectively. (c) Genetic classification of HGNEC is associated with clinical outcome by Kaplan–Meier analysis. (d) Significant difference in copy number of the DEK gene locus between HGNEC-BR1/3 and HGNEC-BR2.

Oncogene DEK oncoprotein in pulmonary endocrine carcinoma T Shibata et al 4673 two groups. In this study, we focused on the DEK gene DEK expression such as that around small vessels, because it encodes a well-known leukemia oncoprotein. which has been suggested to represent a CSC niche in Analysis of other targets (EPS8 encoding an epidermal neural tumors (Calabrese et al., 2007), or a scattered growth factor receptor pathway substrate, and NBAS pattern within the tumor (Figures 2a and c). encoding a neuroblastoma amplified gene and also a fusion partner in acute myeloid leukemia (Fujita et al., 2009)) will be reported separately. We validated the Downregulation of DEK expression in SCLC reduces cell array CGH data for the DEK gene by quantitative PCR proliferation and migration activity in vitro (Pearson’s correlation P ¼ 0.84, Supplementary Figure We examined DEK expression in SCLC cell lines and 2). As shown in Figure 1d, the copy number of the DEK found that a fraction of SCLC cell lines clearly showed gene was significantly higher in tumors in the HGNEC- DEK protein expression (Figure 2f, Supplementary BR1/3 than in those in HGNEC-BR2. Figure 2). To further elucidate the functions of DEK overexpression, we transiently transfected an small hairpin RNA vector targeting the DEK gene into Prognostic significance of DEK overexpression in clinical SBC3 and SBC5 cells to knock down DEK expression, cases of HGNEC both of which are adhesive and show moderate As DEK overexpression has been reported in several expression of DEK protein (Figure 3a). As shown in solid tumors, we investigated DEK protein expression in Figure 3b, we observed that transient knockdown of a larger cohort of HGNEC (79 primary cases) and DEK expression decreased SCLC cell proliferation at attempted to validate its prognostic significance. We the prolonged time (after day 2 point, similar result was observed that DEK protein was concentrated in the observed in N230, a floating SCLC cell line, whereas no nucleus of cancer cells, as well as a subpopulation of growth inhibition was observed in H69 cells, Supple- inflammatory cells (Figures 2a and b). Using the mentary Figure 3 and data not shown). Then we infiltrating lymphocytes as an internal control, we established two independent stable clones (DEKKD-1 segregated the tumors into those showing high expres- and 2) from SBC3 in which DEK expression was sion of DEK (higher than the level in lymphocytes; considerably reduced in comparison with the clone DEK-high, Figure 2c) and those showing equal or weak transfected with the mock vector (Figure 3c). No DEK- DEK expression (DEK-low, Figure 2d). On the basis of downregulated clones were obtained from SBC5 cells. this criterion, we found that patients with DEK-high Similarly in the transient knock down experiments, we tumors (35/79, 44.3% of the total) showed significantly detected a modest reduction of cell proliferation in shorter survival rate than those with DEK-low tumors DEK-silenced clones at day 3 and a gradual increase of (Figure 2e). Thus, DEK expression seemed to be a reduction (about 60% of the control at day 6 point; marker of prognosis in HGNEC. Interestingly, DEK Figure 3d). DEK-silenced clones showed a slight was expressed diffusely in DEK-high tumors, and we decrease in the S-phase fraction and increase in G2/M were unable to discern any characteristic pattern of fraction. There was no increase of the apoptotic fraction

Figure 2 DEK overexpression is a significant prognostic factor of HGNEC. (a) Strong and weak nuclear DEK expression detected in HGNEC cells and surrounding lymphocytes (indicated by L), respectively. (b) Higher magnification of DEK expression in cancer cells. Representative cases of HGNEC (indicated by T) showing strong (c) and weak (d) DEK expression. Bar indicates 100 mm. (e) Significant association between DEK expression and patient survival for HGNEC by Kaplan–Meier analysis. (f) DEK expression in SCLC cell lines. Lane 1: H69, lane2: MS-1, Lane3: N230, lane 4: N417, Lane 5: SBC3, Lane 6: SBC5, Lane 7: Lu130, Lane 8: Lu134A, Lane 9: Lu134B, Lane 10: Lu139, Lane 11: Lu140, Lane 12: Lu165 and Lane 13: PC6. b-actin expression is shown as a loading control. Molecular marker is indicated on the right (kDa).

Oncogene DEK oncoprotein in pulmonary endocrine carcinoma T Shibata et al 4674

Figure 3 DEK downregulation affects SCLC cell proliferation, cell cycle, morphology and migration. (a) Immunoblot analysis of change in DEK expression resulting from transient introduction of control small hairpin RNA (shRNA) and DEK shRNA in SBC3 and SBC5 cells. b-actin expression is shown as a loading control. (b) SBC3 and SBC5 cells were transiently transfected with control shRNA or DEK shRNA, and cell numbers relative to day 0 were plotted for each day. (c) Immunoblot analysis of DEK protein in parent cells (P) and clones stably transfected with control shRNA (Ctrl) and DEK shRNA (DEKKD-1 and 2). (d) Cell numbers relative to day 0 for parent cells (P) and clones stably transfected with control shRNA (Ctrl) and DEK shRNA on each day are plotted (n ¼ 3). (e) Fractions of cells in G1, G2/M and S phases for parent cells (P) and clones stably transfected with control shRNA (Ctrl) and DEK shRNA (n ¼ 3). (f) Cell morphology of clones stably transfected with control shRNA (Control) and DEK shRNA. (g) Numbers of migrated cells in clones stably transfected with control shRNA (Control) and DEK shRNA. Representative area of each transwell is shown at the top (n ¼ 8). Data represents the mean±s.d. Molecular marker is indicated on the right (kDa).

(sub G1) and caspase 3 activation in DEK-silenced cells compared with the control, suggesting that DEK (Figure 3e and Supplementary Figure 4). Morphologi- activity had a role in tumorigenicity of this cell line cally, DEK-silenced clones were flattened compared (Figure 4a). To further test this possibility, we trans- with the control (Figure 3f), and there was a significant planted these clones in immunodeficient mice. Despite of difference in cell motility between DEK-silenced and moderate decrease in in vitro cell proliferation, DEK- control clones (Figure 3g). As these features suggested a silenced clones rarely formed tumors (0/8 and 1/8) cellular senescence-like phenotype, we examined senes- in vivo compared with the parent cells and the control cence-associated b-galactosidase expression in these clone (8/8 and 8/8; Figure 4b). We also tested whether clones. However, we found that DEK-silencing very DEK downregulation has any association with che- rarely induced senescence-associated b-galactosidase motherapy resistance. We compared sensitivity with two positive cells (Supplementary Figure 5). anticancer drugs that are routinely used for treatment of lung cancer, cisplatin and etoposide, between DEK- Reduction of DEK expression robustly decreases in vitro silenced clones and control. Etoposide was recently colony formation, in vivo tumorigenicity, chemotherapy- identified as a selective growth inhibitor of mammary resistance and expression of stem cell markers stem/progenitor cells (Gupta et al., 2009). As in We then performed a replating colony assay to Figure 4c, DEK silencing significantly increased determine the effect of DEK expression silencing on sensitivity to etoposide whereas no significant difference in vitro colony-forming activity. DEK-silenced clones was observed in cisplatin treatment (Supplementary showed a marked reduction in the number of colonies Figure 6). These observations prompted us to evaluate

Oncogene DEK oncoprotein in pulmonary endocrine carcinoma T Shibata et al 4675

Figure 4 DEK downregulation reduces tumorigenicity, lung CSC marker expression and resistance to chemotherapy. (a) Numbers of parental cells, control (Ctrl) and stable DEK-knockdown clones forming colonies in vitro (n ¼ 3). A representative plate showing colony formation by each clone is shown. (b) Numbers of parental cells, control (Ctrl) and stable DEK-knockdown clones forming tumors in vivo (n ¼ 8). (c) Parental cells, control (Ctrl) and stable DEK-knockdown clones were exposed to various concentrations of etoposide for 3 days, and cell numbers relative to day 0 were plotted (n ¼ 3). (d) Quantitative reverse transcriptase PCR analysis of putative stem cell markers in parental cells, control (Ctrl) and stable DEK-knockdown clones. Data represents the mean±s.d.

Table 1 Gene ontology analysis of genes altered by DEK silencing known as Oct-3/4), Sox2, Foxo15 and Klf4), and found GO category Genes in Genes in list P-value that Ascl1, Pou5f1 and its direct target Fbxo15 were category in category specifically downregulated in DEK-silenced clones (Figure 4d and Supplementary Figure 8). GO:8202: steroid metabolism 218 16 0.000000612 GO:16125: sterol metabolism 103 11 0.000000948 GO:6334: nucleosome assembly 118 11 0.00000366 GO:16126: sterol biosynthesis 42 7 0.00000472 DEK maintains expression of transcriptional modifiers GO:31497: chromatin assembly 134 11 0.0000125 that are associated with hematological tumors GO:8610: lipid biosynthesis 326 17 0.0000272 To further understand the global molecular mechanism GO:44255: cellular lipid 712 27 0.0000439 through which DEK regulates the tumorigenicity, we metabolism GO:8203: cholesterol metabolism 95 8 0.000162 compared the genome-wide expression of DEK-silenced GO:6694: steroid biosynthesis 99 8 0.000215 and control clones. Gene ontology analysis revealed GO:6629: lipid metabolism 874 29 0.000235 that the top-ranking molecular pathways altered in GO:6333: chromatin assembly 200 11 0.000457 DEK-silenced clones were sterol/lipid metabolism or disassembly (P ¼ 6.12 Â 10À7) and nucleosome/chromatin assembly GO:6461: protein complex 466 18 0.000676 À6 assembly (P ¼ 3.66 Â 10 ; Table 1). Genes that showed a sig- nificant difference in expression between DEK-silenced and control clones are shown in Table 2 and Supple- mentary Table 2. Among them, we noticed that the the expression of putative lung CSC markers, as CSC expression of a large set of transcriptional regulators has been reported to have an important role in the (such as polycomb complex component, histone modi- efficiency of in vivo tumor formation and drug fier and DNA methyltransferase) and transcriptional resistance. Fluorescence-activated cell sorting analysis factors (especially homeobox and helix-loop-helix types) revealed that SBC3 cells did not express CD133 antigen were altered by DEK silencing. We further validated (Supplementary Figure 7). We also measured Aldh1a1 changes in the expression of these genes by quantitative expression by quantitative reverse transcriptase PCR reverse transcriptase PCR analysis (Figure 5a). Inter- and found that it was significantly decreased in DEK- estingly they included many leukemic translocation silenced clones (Figure 4d). Furthermore, we examined target genes, which may be related to the epigenetic other stem cell-related molecules (Ascl1, Pou5f1 (also maintenance of stem cell features (Table 2).

Oncogene DEK oncoprotein in pulmonary endocrine carcinoma T Shibata et al 4676 Table 2 Representative DEK-regulated genes and their function and relation to cancer Gene name Ratio (Control/DEKshRNA) q-value (%) Function Relation to cancer

Genes downregulated by DEK silencing TLX1 15.2 0 Transcriptional factor, homeobox-type Translocation in leukemia TLE6 4.14 0 Epigenetic regulator Oncogenic in colon cancer HOXA5 3.79 0 Transcriptional factor, homeobox-type Methylated in leukemia LHX9 8.78 0 Transcriptional factor, LIM-homeobox-type CEBPA 2.22 0 Transcriptional factor, bZip-type Translocation in leukemia WNT10B 2.48 0 Growth factor Oncogenic in breast cancer HOXA2 8.37 0 Transcriptional factor, homeobox-type HDAC9 7.94 0 Epigenetic repressor TAL1 3.72 0 Transcriptional factor, HLH-type Translocation in leukemia ABCA1 10.3 1.27 Transporter Drug resistance NEUROD1 3.51 1.55 Transcriptional factor, HLH-type Notch signaling FOXN4 10.4 2.47 Transcriptional factor, forkhead-type DNMT3L 12.3 3.07 DNA methylation FBXO15 2.86 3.24 F-box protein Stem cell DLX2 3.9 3.54 Transcriptional factor, homeobox-type MBD2 3.61 3.54 DNA methylation LHX9 6.63 3.54 Transcriptional factor, homeobox-type FOXJ1 2.66 3.54 Transcriptional factor, forkhead-type TET1 2.86 3.54 Epigenetic regulator Translocation in leukemia DLX1 2.38 3.54 Transcriptional factor, homeobox-type HES1 3.71 3.54 Transcriptional factor, HLH-type Notch signaling DNMT3B 2.51 4.12 DNA methylation DLX4 2.8 4.12 Transcriptional factor, homeobox-type MLLT3 2.45 4.12 Epigenetic regulator, trithorax homolog Translocation in leukemia DLX2 2.7 4.12 Transcriptional factor, homeobox-type NUP98 1.62 4.12 Nuclear pore Translocation in leukemia FOXO4 1.97 4.12 Transcriptional factor, forkhead-type Translocation in leukemia ASCL2 4.5 4.12 Transcriptional factor, HLH-type Oncogenic in colon cancer SETBP1 268 4.46 Epigenetic regulator Translocation in leukemia MED12L 1.67 4.46 Epigenetic regulator

Genes upregulated by DEK silencing NR2F2 0.528 3.44 Nuclear receptor PRDM2 0.563 3.78 Epigenetic regulator Translocation in leukemia AFF1 0.449 3.78 Epigenetic regulator, trithorax homolog Translocation in leukemia HNF4A 0.277 3.78 Transcriptional factor Wnt signaling MYST4 0.554 3.93 Epigenetic regulator Translocation in leukemia FOXN3 0.422 4.43 Transcriptional factor, forkhead-type DNA repair

Abbreviation: HLH, helix-loop-helix.

DEK-associated epigenetic modifier regulates colony Discussion formation activity and lung CSC marker expression To determine whether DEK-regulated epigenetic modi- Molecular classification of pulmonary endocrine tumors fiers are directly implicated in tumor formation activity, In this study, we confirmed that the genetic changes in we selected three candidates (TET1, MLLT3 and TAL1) TC/AC were quite different from those of HGNEC on and knocked down the expression of them using siRNA the basis of the copy number signature, as previous (Supplementary Figure 9). We observed that the studies had suggested (Walch et al., 1998; D’Adda et al., decreased expression of either of them significantly 2005). Our cluster analysis has revealed that only one reduced the replating colony formation activity of SBC3 LCNEC case shared a quite similar genetic profile with cells although it showed little effect on cell proliferation carcinoid tumors, suggesting that malignant progression (Figure 5b and Supplementary Figure 9). Moreover of carcinoid occurs infrequently. Contrary to the downregulation of these DEK-targets also reduced the carcinoid tumor, SCLC and LCNEC showed relatively expression of lung CSC markers (Aldh1a1, Ascl1 and indiscernible copy number profiles, supporting the idea Pou5f1; Figure 5c). Among these candidates, we then that both tumors follow the relatively similar molecular focused on MLLT3 and found that it is frequently carcinogenesis (Jones et al., 2004; Asamura et al., 2006). overexpressed (35/44, 79.5%) and scattered in the Unsupervised classification based on the copy number nucleus of HGNEC cells (Figure 5d). To test whether alterations showed a subtype of HGNEC with a better MLLT3 has any collaborative role with DEK, we prognosis, suggesting the existence of molecular sub- examined the interaction between the two molecules. We types with clinical significance in HGNEC, but this found that MLLT3 coimmunoprecipitated with DEK observation should be validated in a further validation (Figure 5e and Supplementary Figure 10) and further set. Multiple alterations in tumor suppressors and determined the colocalization of MLLT3 and DEK oncogenes have been reported (Sattler and Salgia, proteins in SBC3 cells (Figure 5f). 2003; Righi et al., 2007), however, no prognostic

Oncogene DEK oncoprotein in pulmonary endocrine carcinoma T Shibata et al 4677

Figure 5 DEK-associated epigenetic modifier regulates colony formation activity and lung CSC marker expression. (a) Quantitative RT– PCR analysis of transcriptional modifiers in parental cells (P), control (C) and stable DEK-knockdown clones (KD-1: DEKKD-1, KD-2: DEKKD-2). (b) The numbers of colonies by SBC3 cells transfected with control siRNA (Control) and siRNAs against epigenetic modifiers (MLLT3, TAL1 and TET1) in vitro are shown (n ¼ 3). Representative plate showing colony formation by each cell type is shown below. (c) Quantitative RT–PCR analysis of lung CSC markers in SBC3 cells transfectedwithcontrolsiRNAsand siRNAs against epigenetic modifiers. (d) Representative immunohistochemistry for MLLT3 in a case of primary HGNEC. Bar indicates 100 mm. (e) Immuno- precipitates obtained with anti-DEK antibody from SBC3 cells were immunoblotted with anti-MLLT3 (upper) and anti-DEK (lower) antibodies. Asterisk indicates immunoglobulin. (f) Immunofluorescent analysis revealed colocalization of DEK (red) and MTTL3 (green) proteins in SBC3 cells. Bar indicates 10 mm. Data represents the mean±s.d. Molecular markers are indicated on the right (kDa). biomarker has been currently reported in HGNEC and was significantly associated with a shorter time to our study has also failed to correlate our molecular recurrence in a large set of clinical samples. classification with known oncoproteins. Therefore we further extracted specific loci that were associated with the prognostic classification to identify a novel diag- DEK is a putative oncoprotein associated with nostic biomarker. We focused on one of the targets, the tumor-initiating activity and regulates the DEK gene, and quantitative copy number analyses transcriptional modifiers expression confirmed that the increased DEK gene dose was DEK encodes a nuclear protein that has been reported significantly associated with poor prognosis. to exhibit a wide range of nuclear function; it modifies Several previous studies have reported that increased the topological conformation of DNA at replication or DEK expression is associated with malignant features in transcription (Alexiadis et al., 2000; Gamble and Fisher, liver and colon cancers and melanoma (Kondoh et al., 2007), affects histone acetylation and the RNA splicing 1999; Han et al., 2009; Khodadoust et al., 2009). machinery (Ko et al., 2006; Soares et al., 2006), Although it has been reported that HGNEC has a very positively and negatively regulates transcription (Fu poor clinical outcome (Asamura et al., 2006; Gustafsson et al., 1997; Campillos et al., 2003; Sammons et al., et al., 2008), previous studies have suggested the 2006), and is implicated in DNA damage response existence of subgroups that differ in their clinical courses (Kappes et al., 2008). DEK expression is regulated by (Jones et al., 2004). Consistent with the genetic and E2F and overexpression of DEK protein has been functional data, we confirmed that DEK overexpression reported in several types of solid tumor (Carro et al.,

Oncogene DEK oncoprotein in pulmonary endocrine carcinoma T Shibata et al 4678 2006). Moreover, DEK inhibits apoptosis by modulat- such as the trithorax complex (MLLT2 and MLLT3), ing p53 activity and is implicated in cellular senescence/ histone acetyltransferase (MYST4), de novo DNA immortalization (Wise-Draper et al., 2006; Scoumanne methyltransferase and methyl-DNA binding protein and Chen, 2007). It has been reported that DEK has (DNMT3B, DNMT3L and MBD2), 5-hydroxymethyl- transforming activity in human keratinocytes when cytosine converting enzyme (TET1), transcriptional working in association with HPV E6/E7 viral oncopro- activator–repressor complex (MED12L, HDAC9, teins or p16 silencing (Wise-Draper et al., 2005, 2009a). SETBP1 and PRDM2), and several transcriptional A large-scale resequence project of cancer genome has factors (especially, homeobox and helix-loop-helix-type revealed a heterozygous missense mutation (K348N) of DNA binding molecules). Surprisingly, these include DEK in a case of renal tumor (http://www.sanger.ac.uk/ many leukemia-associated oncoproteins that are part- perl/genetics/CGP/cosmic?action ¼ sample;id ¼ 948640). ners of disease-causing interchromosomal transloca- Thus, DEK itself can function as an oncoprotein in solid tions. As epigenetic regulation is a crucial factor for tumors. In this study, we detected frequent gain and establishment and maintenance of normal stem cells as amplification of the DEK gene in a subset of HGNEC well as CSC (Glinsky, 2008), it can be hypothesized that with poor prognosis. Delayed and moderate decrease of these oncoproteins may have a pivotal role in the cell proliferation and marked reduction of in vitro epigenetic signature of CSC in both hematopoietic and colony formation activity were observed when DEK was pulmonary endocrine tumors, and we further showed silenced in SCLC cell lines. However, no significant that downregulation of these molecules affected the increase of apoptosis or senescence was detected in colony formation activity and the expression of CSC contrast to previous reports of other cell types, whereas markers. Thus, DEK may maintain the tumor-initiating p53 expression was induced in SBC3 cells (Supplemen- activity of solid tumors, at least partly through tary Figure 4, Wise-Draper et al., 2006, 2009a; modulation of these transcriptional regulators, suggest- Khodadoust et al., 2009). This might be because both ing that a common molecular pathway functions in the the RB and p53 pathways are frequently inactivated and maintenance of both liquid and solid CSCs. It has been could partly complement DEK-mediated anti-apoptotic shown that DEK overexpression interferes with differ- and anti-senescence function in SCLC cell lines. More- entiation and expands the population of undifferen- over, we have revealed that DEK has an important role tiated keratinocytes (Wise-Draper et al., 2009b), in maintaining the tumorigenicity of SCLC cell lines supporting the contention that DEK also has a in vivo, again supporting the possibility that DEK significant role in normal epithelial stem cell main- functions as an oncoprotein in HGNEC, whereas tenance. complementary proof by DEK overexpression could It remains unknown how much population of not be obtained because of the low transfection HGNEC is contributing to tumor-initiating in vivo.As efficiency in SCLC cell lines (data not shown). DEK is widely expressed in primary HGNEC, an arising As the major phenotypes associated with DEK question is how DEK specifically regulates a set of genes downregulation are in vivo tumor formation, chemother- in distinct cellular contexts, such as CSC. One possibility apy-resistance and poor prognosis, we hypothesized that is that additional modification of DEK might determine DEK has a role in tumor-initiating or CSC activity. A its target specificity (Kappes et al., 2004; Scoumanne basic helix-loop-helix-type transcriptional factor, and Chen, 2007). Another, but not exclusive, idea is that achaete–scute complex homologue 1, has been shown additional cooperation with other transcriptional fac- to regulate the tumor-initiating activity of SCLC (Jiang tors or nucleosome/chromatin assembly factors may et al., 2009), and we found that the expression of lung mark specific DEK binding or work cooperatively and CSC markers (Aldh1a1 and Ascl1) and putative induce characteristic transcriptional changes, as DEK embryonic stem cell and somatic reprogramming master has been shown to interact with several epigenetic regulators (Pou5f1/Oct4 and its direct target Fbxo15; modifiers (Hollenbach et al., 2002; Cleary et al., 2005; Tokuzawa et al., 2003; Okita et al., 2007) were down- Cavella´n et al., 2006). These hypotheses may also regulated by silencing of DEK in SCLC cell lines, explain why broadly expressed DEK is associated with supporting the contention that DEK maintains CSC the tumor-initiating cells, a supposedly rare tumor features in HGNEC. As it has been shown that Notch subpopulation. That is, DEK overexpression may be signaling has a role in maintenance of neural stem cell required, but not sufficient, for establishment of tumor- and also in SCLC (Sriuranpong et al., 2001; Mizutani initiating activity, and further modification or recruit- et al., 2007), and we detected altered expression of ment of associated factors might be necessary for its full several Notch signal-associated genes (Table 2) by DEK oncogenic activity. We tested the latter possibility and downregulation, we examined activation of Notch focused on MLLT3 because it functions as a transcrip- signaling in DEK-silenced clones. However, we did not tional activator itself in a similar way as the polycomb observe any significant change in CBF1-dependent complex, which is known to be associated with CSC transcriptional activity, which is activated by Notch (Schuettengruber et al., 2007; Widschwendter et al., signaling (Hsieh et al., 1997, Supplementary Figure 11). 2007), and directly targets TAL1 and the Hox gene To elucidate the molecular roles of DEK in these cluster (Collins et al., 2002; Pina et al., 2008), which are phenotypes, we conducted expression-profiling analysis. also downregulated in DEK-silenced clones, and On the basis of genome-wide expression analysis, we MLLT3 downregulation showed little effect on in vitro found that DEK regulates transcriptional regulators, SCLC cell proliferation. We found that MLLT3

Oncogene DEK oncoprotein in pulmonary endocrine carcinoma T Shibata et al 4679 associates directly or indirectly with DEK and partially previously (Shibata et al., 2003, 2008). For cell sorting analysis, rescued colony-forming activity in DEK-silenced cells cells were stained with propidium iodide (Sigma-Aldrich, St Louis, (Supplementary Figure 9). MLLT3 translocation has a MO, USA) for cell cycle analysis, or anti-CD133 antibody major role in leukemic stem cell development and is (CD133/1-PE, Miltenyi Biotec, Bergisch Gladbach, Germany) tethered to the potential CSC-related genes (Krivtsov for lung cancer stem cell analysis. Fluorescence-activated cell et al., 2006). Our preliminary study suggested that DEK sorting analysis was performed using fluorescence-activated cell sorting Calibur (BD Biosciences, San Jose, CA, USA) as indirectly associates with MLLT3, and the MLLT3– the manufacture’s protocol. For protein extraction, we used DEK nuclear complex might be recruited to the genes a slightly modified buffer (10 mM Tris-HCl (pH7.5), 175 mM regulating lung tumor-initiating activity. Identification NaCl, 5 mM EDTA, 0.5% Triton-X, 0.5% NP-40 with a of the molecular mechanism underlying CSC mainte- proteinase inhibitor cocktail (Roche, Mannheim, Germany), nance is critical for devising new CSC-targeting thera- and the procedures of immunoprecipitation and immunoblot- pies in lung cancer and further characterization of the ting were preformed as described (Shibata et al., 2008). specific nuclear DEK complex in CSC will be significant. For assessment of in vivo tumorigenicity, 1 Â 106 cells were In conclusion, our study has proposed that HGNEC subcutaneously transplanted into the trunks of nude mice. The can be genetically classified and identification of novel mice were kept at the Animal Care and Use Facilities at the molecular signatures should be valuable for the mole- National Cancer Center under specific pathogen-free condi- tions and all experiments were approved by the Institutional cular diagnosis and individualized therapy of this Animal Care and Ethics Committee. malignant tumor. Our results also suggested that DEK has a direct role in determining the malignant features and tumor-initiating activity of HGNEC and could be a Gene expression analysis and quantitative PCR potential biomarker for assessment of prognosis and an In all, 10mg of total RNA was reverse-transcribed by MMLV reverse transcriptase, and a Cy3-labeled cRNA probe was attractive target of extensive therapeutic application synthesized using T7 RNA polymerase. Probes were hybri- including elimination of CSC population. dized with a microarray containing 41 000 long oligonucleo- tides covering the whole human genome (Whole Human Genome Oligo Microarray, G4112F, Agilent Technologies, Materials and methods Santa Clara, CA, USA). After washing, the microarray was scanned by the DNA microarray scanner (Agilent Techno- Details of the experimental procedures are described in the logies). Quantitative genomic PCR and reverse transcriptase Supplementary Material. PCR was performed in triplicate and evaluated using universal probes for each amplicon and the Light-cycler system (Roche). Primers designed by ProbeFinder (Version 2.45, Roche) and Clinical samples, DNA extraction and immunohistochemistry used in this study are shown in Supplementary Table 3. The Surgical specimens from 101 Japanese patients with lung relative expression of each gene was determined by comparison neuroendocrine tumor (11 TCs, 2 ACs, 25 SCLCs and 63 with that of glyceraldehyde 3-phosphate dehydrogenase. LCNECs), who were diagnosed and underwent surgery at the National Cancer Center Hospital, Tokyo, Japan, between March 1982 and January 2004 were examined. Clinicopatho- Statistics logical data of the analyzed cases are shown in the Array CGH and microarray raw-data analysis was first Supplementary Table 4. Only cases with a consensus histo- conducted using the Gene Spring software package (Agilent logical diagnosis agreed by the central board of pathologists technologies). Two-dimensional hierarchical clustering analy- were analyzed (Asamura et al., 2006). Fragments of tumor sis of the samples and signal ratios was performed using the were fixed with 100% methanol and embedded in paraffin. To Impressionist (Gene Data, Basel, Switzerland) and GeneMaths obtain pure tumor DNA from 49 cases for array CGH (Applied Maths, Sint-Martens-Latem, Belgium) software analysis, laser-capture microdissection using an LM200 (MDS programs, as described previously (Shibata et al., 2005; Katoh Analytical Technologies, Toronto, Canada) and the whole- et al., 2007). As we hypothesized that array CGH data includes genome PCR method were conducted. Immunohistochemistry the genetic heterogeneity of cancer cells that frequently occurs was performed as previously described (Shibata et al., 2005). in primary tumor, we took the strategy for discovering This study protocol for clinical samples was approved by the prognostic BAC clones as follows: we directly used normalized institutional review board of the National Cancer Center. ratio data for assessing prognostic BAC clones in the first set of samples, and then validate the results by an independent immunohistochemical analysis of the second sample set. We Array CGH analysis used the Significance Analysis of Microarrays software For array CGH analysis, a custom-made focused CGH array, package (SAM; http://www-stat.stanford.edu/Btibs/SAM/ consisting of 800 duplicated BAC clones corresponding to index.html) to select prognostic BAC clones and to identify various chromosomal loci that are altered in various human genes differentially expressed between control and DEK- cancers, was used (Sonoda et al., 2004). Array CGH data of 23 silenced clones. After 100 rounds of permutation test, an HGNEC samples (8 SCLC and 15 LCNEC cases) have been adjusted P-value and a false discovery rate were estimated. The previously reported (Peng et al., 2005). Sixteen-bit fluorescence Kaplan–Meier method was used to estimate the probability of intensity TIF images were obtained using a scanner (FLA8000, disease-free survival. Fuji Film, Tokyo, Japan) and analyzed using GenePix Pro 5.0 (Axon Instruments Inc., Foster City, CA, USA).

Cell biological and biochemical experiments Conflict of interest Cell proliferation, cell migration assay, soft agar colony assay and immunofluorescent analysis were performed as described The authors declare no conflict of interest.

Oncogene DEK oncoprotein in pulmonary endocrine carcinoma T Shibata et al 4680 Acknowledgements was supported in part by Grant-in-Aid for Cancer Research from the Ministry of Health, Labor and Welfare of Japan, a We thank Dr Weixia Peng for the initial work of array CGH grant from the New Energy and Industrial Technology and Dr Yukihiro Yoshida for the help of clinical data analysis. Development Organization (NEDO), Japan and a grant for We thank Dr S Diane Hayward at the Johns Hopkins the program for promotion of Fundamental Studies in Health University for generously providing the CBF1 reporter Sciences of the National Institute of Biomedical Innovation constructs (4xwtCBF1Luc and 4xmtCBF1Luc). This work (NiBio).

References

Alexiadis V, Waldmann T, Andersen J, Mann M, Knippers R, Gruss Hsieh JJ, Nofziger DE, Weinmaster G, Hayward SD. (1997). Epstein- C. (2000). The protein encoded by the proto-oncogene DEK Barr virus immortalization: Notch2 interacts with CBF1 and blocks changes the topology of chromatin and reduces the efficiency of differentiation. J Virol 71: 1938–1945. DNA replication in a chromatin-specific manner. Genes Dev 14: Jiang T, Collins BJ, Jin N, Watkins DN, Brock MV, Matsui W et al. 1308–1312. (2009). Achaete-scute complex homologue 1 regulates tumor- Asamura H, Kameya T, Matsuno Y, Noguchi M, Tada H, Ishikawa Y initiating capacity in human small cell lung cancer. Cancer Res 69: et al. (2006). Neuroendocrine neoplasms of the lung: a prognostic 845–854. spectrum. J Clin Oncol 24: 70–76. Jones MH, Virtanen C, Honjoh D, Miyoshi T, Satoh Y, Okumura S Calabrese C, Poppleton H, Kocak M, Hogg TL, Fuller C, Hamner B et al. (2004). Two prognostically significant subtypes of high-grade et al. (2007). A perivascular niche for brain tumor stem cells. Cancer lung neuroendocrine tumours independent of small-cell and large- Cell 11: 69–82. cell neuroendocrine carcinomas identified by gene expression Campillos M, Garcı´a MA, Valdivieso F, Va´zquez J. (2003). profiles. Lancet 363: 775–781. Transcriptional activation by AP-2alpha is modulated by the Kappes F, Damoc C, Knippers R, Przybylski M, Pinna LA, Gruss C. oncogene DEK. Nucleic Acids Res 31: 1571–1575. (2004). Phosphorylation by protein kinase CK2 changes the DNA Carro MS, Spiga FM, Quarto M, Di Ninni V, Volorio S, Alcalay M binding properties of the human chromatin protein DEK. Mol Cell et al. (2006). DEK Expression is controlled by E2F and deregulated Biol 24: 6011–6020. in diverse tumor types. Cell Cycle 5: 1202–1207. Kappes F, Fahrer J, Khodadoust MS, Tabbert A, Strasser C, Mor- Cavella´n E, Asp P, Percipalle P, Farrants AK. (2006). The WSTF- Vaknin N et al. (2008). DEK is a poly(ADP-ribose) acceptor in SNF2h chromatin remodeling complex interacts with several apoptosis and mediates resistance to genotoxic stress. Mol Cell Biol nuclear proteins in transcription. J Biol Chem 281: 16264–16271. 28: 3245–3257. Cleary J, Sitwala KV, Khodadoust MS, Kwok RP, Mor-Vaknin N, Katoh H, Ojima H, Kokubu A, Saito S, Kondo T, Kosuge T et al. Cebrat M et al. (2005). p300/CBP-associated factor drives DEK into (2007). Genetically distinct and clinically relevant classification of interchromatin granule clusters. J Biol Chem 280: 31760–31767. hepatocellular carcinoma: putative therapeutic targets. Gastroenter- Collins EC, Appert A, Ariza-McNaughton L, Pannell R, Yamada Y, ology 133: 1475–1486. Rabbitts TH. (2002). Mouse Af9 is a controller of embryo Khodadoust MS, Verhaegen M, Kappes F, Riveiro-Falkenbach E, patterning, like Mll, whose human homologue fuses with Af9 after Cigudosa JC, Kim DS et al. (2009). Melanoma proliferation and chromosomal translocation in leukemia. MolCellBiol22: 7313–7324. chemoresistance controlled by the DEK oncogene. Cancer Res 69: D’Adda T, Bottarelli L, Azzoni C, Pizzi S, Bongiovanni M, Papotti M 6405–6413. et al. (2005). Malignancy-associated X chromosome allelic losses in Ko SI, Lee IS, Kim JY, Kim SM, Kim DW, Lee KS et al. (2006). foregut endocrine neoplasms: further evidence from lung tumors. Regulation of histone acetyltransferase activity of p300 and PCAF Mod Pathol 18: 795–805. by proto-oncogene protein DEK. FEBS Lett 580: 3217–3222. Fu GK, Grosveld G, Markovitz DM. (1997). DEK, an autoantigen Kondoh N, Wakatsuki T, Ryo A, Hada A, Aihara T, Horiuchi S et al. involved in a chromosomal translocation in acute myelogenous (1999). Identification and characterization of genes associated leukemia, binds to the HIV-2 enhancer. Proc Natl Acad Sci USA 94: with human hepatocellular carcinogenesis. Cancer Res 59: 1811–1815. 4990–4996. Fujita K, Sanada M, Harada H, Mori H, Niikura H, Omine M et al. Krivtsov AV, Twomey D, Feng Z, Stubbs MC, Wang Y, Faber J et al. (2009). Molecular cloning of t(2;7)(p24.3;p14.2), a novel chromo- (2006). Transformation from committed progenitor to leukaemia somal translocation in myelodysplastic syndrome-derived acute stem cell initiated by MLL-AF9. Nature 442: 818–822. myeloid leukemia. J Hum Genet 54: 355–359. Mizutani K, Yoon K, Dang L, Tokunaga A, Gaiano N. (2007). Gamble MJ, Fisher RP. (2007). SET and PARP1 remove DEK from Differential Notch signalling distinguishes neural stem cells from chromatin to permit access by the transcription machinery. Nat intermediate progenitors. Nature 449: 351–355. Struct Mol Biol 14: 548–555. Okita K, Ichisaka T, Yamanaka S. (2007). Generation of germline- Glinsky GV. (2008). ‘Stemness’ genomics law governs clinical behavior competent induced pluripotent stem cells. Nature 448: 313–317. of human cancer: implications for decision making in disease Pina C, May G, Soneji S, Hong D, Enver T. (2008). MLLT3 regulates management. J Clin Oncol 26: 2846–2853. early human erythroid and megakaryocytic cell fate. Cell Stem Cell Gupta PB, Onder TT, Jiang G, Tao K, Kuperwasser C, Weinberg RA 2: 264–273. et al. (2009). Identification of selective inhibitors of cancer stem cells Peacock CD, Watkins DN. (2008). Cancer stem cells and the ontogeny by high-throughput screening. Cell 138: 645–659. of lung cancer. J Clin Oncol 26: 2883–2889. Gustafsson BI, Kidd M, Chan A, Malfertheiner MV, Modlin IM. Peng WX, Shibata T, Katoh H, Kokubu A, Matsuno Y, Asamura H (2008). Bronchopulmonary neuroendocrine tumors. Cancer 113: 5–21. et al. (2005). Array-based comparative genomic hybridization Han S, Xuan Y, Liu S, Zhang M, Jin D, Jin R et al. (2009). analysis of high-grade neuroendocrine tumors of the lung. Cancer Clinicopathological significance of DEK overexpression in serous Sci 96: 661–667. ovarian tumors. Pathol Int 59: 443–447. Righi L, Volante M, Rapa I, Scagliotti GV, Papotti M. (2007). Neuro- Hollenbach AD, McPherson CJ, Mientjes EJ, Iyengar R, Grosveld G. endocrine tumours of the lung. A review of relevant pathological (2002). Daxx and histone deacetylase II associate with chromatin and molecular data. Virchows Arch 451(Suppl 1): S51–S59. through an interaction with core histones and the chromatin- Sammons M, Wan SS, Vogel NL, Mientjes EJ, Grosveld G, Ashburner associated protein Dek. J Cell Sci 115: 3319–3330. BP. (2006). Negative regulation of the RelA/p65 transactivation

Oncogene DEK oncoprotein in pulmonary endocrine carcinoma T Shibata et al 4681 function by the product of the DEK proto-oncogene. J Biol Chem Travis WD, Linnoila RI, Tsokos MG, Hitchcock CL, Cutler Jr GB, 281: 26802–26812. Nieman L et al. (1991). Neuroendocrine tumors of the lung with Sattler M, Salgia R. (2003). Molecular and cellular biology of small proposed criteria for large-cell neuroendocrine carcinoma: an cell lung cancer. Semin Oncol 30: 57–71. ultrastructural, immunohistochemical, and flow cytometric study Schuettengruber B, Chourrout D, Vervoort M, Leblanc B, Cavalli G. of 35 cases. Am J Surg Pathol 15: 529–553. (2007). Genome regulation by polycomb and trithorax proteins. Cell Visvader JE, Lindeman GJ. (2008). Cancer stem cells in solid tumours: 128: 735–745. accumulating evidence and unresolved questions. Nat Rev Cancer 8: Scoumanne A, Chen X. (2007). The lysine-specific demethylase 1 755–768. is required for cell proliferation in both p53-dependent and von Lindern M, Fornerod M, van Baal S, Jaegle M, de Wit T, -independent manners. J Biol Chem 282: 15471–15475. Buijs A et al. (1992). The translocation (6;9), associated Shibata T, Chuma M, Kokubu A, Sakamoto M, Hirohashi S. (2003). with a specific subtype of acute myeloid leukemia, results in the EBP50, a beta-catenin-associating protein, enhances Wnt signaling and fusion of two genes, dek and can, and the expression of a is over-expressed in hepatocellular carcinoma. Hepatology 38: 178–186. chimeric, leukemia-specific dek-can mRNA. Mol Cell Biol 12: Shibata T, Kokubu A, Gotoh M, Ojima H, Ohta T, Yamamoto M 1687–1697. et al. (2008). Genetic alteration of Keap1 confers constitutive Nrf2 Walch AK, Zitzelsberger HF, Aubele MM, Mattis AE, Bauchinger M, activation and resistance to chemotherapy in gallbladder cancer. Candidus S et al. (1998). Typical and atypical carcinoid tumors of Gastroenterology 135: 1358–1368. the lung are characterized by 11q deletions as detected by Shibata T, Uryu S, Kokubu A, Hosoda F, Ohki M, Sakiyama T et al. comparative genomic hybridization. Am J Pathol 153: 1089–1098. (2005). Genetic classification of lung adenocarcinoma based on Widschwendter M, Fiegl H, Egle D, Mueller-Holzner E, Spizzo G, array-based comparative genomic hybridization analysis: its associa- Marth C et al. (2007). Epigenetic stem cell signature in cancer. Nat tion with clinicopathologic features. Clin Cancer Res 11: 6177–6185. Genet 39: 157–158. Sriuranpong V, Borges MW, Ravi RK, Arnold DR, Nelkin BD, Wise-Draper TM, Allen HV, Jones EE, Habash KB, Matsuo H, Baylin SB et al. (2001). Notch signaling induces cell cycle arrest in Wells SI. (2006). Apoptosis inhibition by the human DEK small cell lung cancer cells. Cancer Res 61: 3200–3205. oncoprotein involves interference with p53 functions. Mol Cell Biol Soares LM, Zanier K, Mackereth C, Sattler M, Valca´rcel J. (2006). 26: 7506–7519. Intron removal requires proofreading of U2AF/30 splice site Wise-Draper TM, Allen HV, Thobe MN, Jones EE, Habash KB, recognition by DEK. Science 312: 1961–1965. Mu¨nger K et al. (2005). The human DEK proto-oncogene is a Sonoda I, Imoto I, Inoue J, Shibata T, Shimada Y, Chin K et al. (2004). senescence inhibitor and an upregulated target of high-risk human Frequent silencing of low density lipoprotein receptor-related protein papillomavirus E7. J Virol 79: 14309–14317. 1B (LRP1B) expression by genetic and epigenetic mechanisms in Wise-Draper TM, Mintz-Cole RA, Morris TA, Simpson DS, esophageal squamous cell carcinoma. Cancer Res 64: 3741–3747. Wikenheiser-Brokamp KA, Currier MA et al. (2009a). Over- Tokuzawa Y, Kaiho E, Maruyama M, Takahashi K, Mitsui K, Maeda expression of the cellular DEK protein promotes epithelial M et al. (2003). Fbx15 is a novel target of Oct3/4 but is dispensable transformation in vitro and in vivo. Cancer Res 69: 1792–1799. for embryonic stem cell self-renewal and mouse development. Mol Wise-Draper TM, Morreale RJ, Morris TA, Mintz-Cole RA, Hoskins Cell Biol 23: 2699–2708. EE, Balsitis SJ et al. (2009b). DEK proto-oncogene expression Travis WD, Brambilla E. (2004). Pathology and Genetics of Tumors of interferes with the normal epithelial differentiation program. Am the Lung, Pleura, Thymus and Heart. IARC Press: Lyon, pp 341. J Pathol 174: 71–81.

Supplementary Information accompanies the paper on the Oncogene website (http://www.nature.com/onc)

Oncogene