sign in sign up
<<
Home , ATBF1, Engrailed (gene), HOXA9, ID4, KLF2, NFX1

(2007) 26, 3352–3363 & 2007 Nature Publishing Group All rights reserved 0950-9232/07 $30.00 www.nature.com/onc ORIGINAL ARTICLE Combined effects of the two reciprocal t(4;11)fusion MLL . AF4 and AF4 . MLL confer resistance to , cell cycling capacity and growth transformation

A Gaussmann1, T Wenger2,3, I Eberle1, A Bursen1, S Bracharz1, I Herr2, T Dingermann1 and R Marschalek1

1Institute of Pharmaceutical Biology/ZAFES, JWG-University Frankfurt, Biocenter, Frankfurt/Main, Germany and 2Molecular OncoSurgery, Department of Surgery, University of Heidelberg and German Research Center, Heidelberg, Germany

The reciprocal chromosomal translocation t(4;11)is Introduction correlated with infant, childhood, adult and therapy- related high-risk acute leukemia. Here, we investigated Chromosomal translocations t(4;11) are the most the biological effects of MLL . AF4, AF4 . MLL or the frequent chromosomal aberration of the MLL combination of both reciprocal fusion proteins in a and are correlated with infant, childhood, adult and conditional in vitro cell culture model system. Several therapy-related acute lymphoblastic leukemia (Felix parameters like cell growth, cell cycling capacity, et al., 1995; for a review see Mitterbauer-Hohendanner apoptotic behavior and growth transformation were and Mannhalter, 2004). In particular, t(4;11) patients investigated under physiological and stress conditions. have a poor prednisone response (Dordelmann et al., Co-transfected cells displayed the highest resistance 1999) and are therefore stratified according to high-risk against apoptotic triggers, cell cycling capacity and loss- acute leukemia protocols (Pui et al., 1994). of-contact inhibition. These analyses were complemented Although t(4;11) translocations are associated with by profiling experiments and specific gene such a severe disease , no animal model is signatures were established for each of the three cell lines. currently available that phenocopies the human disease. Interestingly, co-transfected cells strongly upregulate the By using the inverter mouse model and an AF4 knock-in gene Nanog. In combination with Oct4, the strategy, however, two different transgenic MLL . AF4 Nanog homeoprotein is steering maintenance of pluripo- mouse models have been established recently. Both tency and self-renewal in embryonic stem cells. Transcrip- developed a disseminated large B-cell lymphoma with tion of Nanog and other factors, like Oct4 and low penetrance and long latency: about 50% of these Bmi1, was verified in biopsy material of t(4;11)patient mice died of B-cell lymphomas after a median time of cells which express both reciprocal t(4;11)fusion . In 520–540 days (Chen et al., 2006; Metzler et al., 2006). conclusion, the presence of both reciprocal MLL fusion This clearly indicated that the MLL . AF4 fusion proteins confers biological properties known from t(4;11) alone does not cause the known t(4;11) disease leukemia, suggesting that each of the two fusion proteins phenotype. Besides these transgenic attempts, retroviral contribute specific properties and, in combination, also transduction of the MLL . AF4 fusion gene does not lead synergistic effects to the leukemic phenotype. to the development of any hematomalignancy (So et al., Oncogene (2007) 26, 3352–3363. doi:10.1038/sj.onc.1210125; 2004; Lavau et al., 2004, personal communication), even published online 27 November 2006 though a small number of blast-like replating cells could be obtained in semisolid agar. Keywords: MLL; AF4; acute leukemia; t(4;11) fusion Apart from these animal model systems, only a few proteins; Nanog stable cell lines were established from t(4;11) patients; however, they all differ in their biological properties and their karyotypes (Stong et al., 1985; Lange et al., 1987; Greil et al., 1994). They are characterized by the expression of lymphatic and myeloid surface antigens (e.g. CD34 þ , CD19 þ , CD13 þ , CD33 þ , CD133 þ , CD10À) and their ability to efficiently blockapoptosis Correspondence: Professor R Marschalek, Institute of Pharmaceutical under cytotoxic drug treatment (Kersey et al., 1998). Biology/ZAFES, University of Frankfurt, Biocenter, Max-von-Laue- This biological feature may explain the high relapse rate Str. 9, D-60439 Frankfurt/Main, Germany. in these leukemia patients. By contrast, overexpression E-mail: [email protected] of the MLL . AF4 fusion protein in the myelomonocytic 3Present address: Centre d’Immunologie de Marseille-Luminy, Mar- seille, France. leukemia cell line U937 cells caused arrest Received 11 July 2006; revised 2 October 2006; accepted 13 October (Caslini et al., 2004), whereas siRNA-mediated knock- 2006; published online 27 November 2006 down of the MLL . AF4 fusion protein in the t(4;11) t(4;11) pathobiology A Gaussmann et al 3353 leukemic cell line SEM led to a strong increase of apoptosis and a reduction in clonogenicity (Thomas et al., 2005). Overexpression of the reciprocal AF4 . MLL fusion protein in MEF led to a loss-of- contact inhibition, and thus to growth transformation (Bursen et al., 2004). All these data sound controversial and did not permit consistent conclusions to be drawn on the pathological disease mechanism(s) of t(4;11) leukemia. Therefore, we aimed to establish an in vitro t(4;11) model system that recapitulates certain aspects of Figure 1 Doxycycline-dependent expression of MLL fusion genes t(4;11) leukemia cells, and thus allows investigation of and protein expression in stably transfected cells. (a) RT–PCR experiments. M ¼ size marker; A ¼ AF4 . MLL-transfected cell line; the specific contribution(s) of each reciprocal MLL B ¼ MLL . AF4-transfected cell line; C ¼ co-transfected cell line; fusion protein. N ¼ negative (water) control; P ¼ positive control (cloned cDNA); For this purpose, mammalian cells were stably 7Dox ¼ presence or absence of 10 mg/ml doxycycline. (b) Western transfected with conditional expression constructs cod- blot experiments. Lane 1: untransfected MEF/tTA cell line; lanes . 2–5: MEF/tTA cell lines stably transfected with Tet-off expression ing for doxycycline-inducible MLL AF4 and/or constructs coding for AF4 . MLL (der4), MLL . AF4 (der11) or AF4 . MLL transgenes. After induction, transfected cells both (der4/11); protein expression was induced by omitting were analysed for their growth properties (cell counts doxycycline in the growth media for 5 days. Upper panel: and 3-[4,5-dimethylthiazol-2-yl]-2,5-diphenyl tetrazo- expression of the AF4 . MLL fusion protein; middle panel: lium bromide (MTT) ), cell cycling capacity, expression of the MLL . AF4 fusion protein; lower panel: b-actin loading controls. Transfected cell lines were treated with MG132 to specific apoptotic rates and growth transformation. In stabilize the expression of the full-length AF4 . MLL fusion protein parallel, gene expression profiling experiments were (328 kDa) that otherwise would be degraded via the proteasomal performed to gain insights into up- and downregulated pathways due to its ability to interact with the E3-ligases SIAH1 or genes due to the presence of either one or both SIAH2, respectively (Bursen et al., 2004). reciprocal MLL fusion proteins. The results obtained suggest that both reciprocal MLL fusion proteins are contributing to the disease phenotype of t(4;11) leuke- mic cells. Moreover, gene expression profiling experi- expression of MLL . AF4 and AF4 . MLL in all three ments revealed the transcriptional upregulation of clones used throughout this study. Nanog and Oct4 in co-transfected cells, whereas single- transfected cells did not show any upregulation of these genes. This novel feature was verified in the biopsy Cell growth, specific apoptosis and cell cycle analysis of material of t(4;11) patients. Nanog codes for an single- and co-transfected cells under normal conditions -class homeobox protein which is – in and serum deprivation conjunction with Oct4 and – necessary and All cell clones were grown in the absence of doxycycline sufficient for the self-renewal and maintenance of for 5 days and investigated under normal cell culture embryonic stem cells (Chambers et al., 2003; Mitsui conditions (Dulbecco’s modified Eagle’s medium et al., 2003). The consequences of these novel findings (DMEM)/10%fetal calf serum (FCS)/1% Pen-Strep). will be discussed. A significantly higher growth rate was observed for co- transfected cells, followed by untransfected MEF/tTA cells and AF4 . MLL-transfected cells. Nearly no growth Results was observed for MLL . AF4-transfected cells, indicating that these cells are inhibited in their growth potential Establishing conditional cell lines expressing MLL . AF4, (Figure 2a, see below). Noteworthy, AF4 . MLL-trans- AF4 . MLL or both MLL fusion genes fected and co-transfected cells displayed significantly Stably transfected MEF/tTA cell lines (Tet-off) were higher growth rates according to the data obtained by generated by transfecting them with the above men- the MTT assay, but the net growth of AF4 . MLL- tioned expression constructs. Single-cell clones were transfected cells seems to be compromised due to a selected under appropriate conditions and tested for higher apoptosis rate (2–20% over the observation their ability to induce their transgene(s) (data not period; see below). All other cell lines displayed an shown). Inducible cell clones were chosen and used apoptosis rate below 3% under these cell culture throughout the experiments. As shown in Figure 1a, a conditions (Figure 2b). The MTT assays revealed that doxycycline-dependent increase in could the MEF/tTAHder11 cells displayed significantly lower be demonstrated for all transgenes after withdrawal of growth rates than the MEF/tTAHder4 and the MEF/ doxycycline (À). Time course experiments revealed tTAHder4/11 cell lines, respectively. It was also obvious expression of the resulting fusion proteins after 3–5 that untransfected control cells and MLL . AF4-trans- days (data not shown). Therefore, all subsequent fected cells seem to reduce their growth rates when experiments were performed after 5 days of transgene confluency was reached, whereas AF4 . MLL- and co- induction in order to generate consistent data sets. transfected cells displayed increasing growth rates which Expression of both MLL fusion proteins at day 5 is reflect their ‘loss-of-contact inhibition’ (Figure 2c; see shown in Figure 1b to demonstrate full-length protein also below: Focus formation experiments).

Oncogene t(4;11) pathobiology A Gaussmann et al 3354

Figure 2 Cell growth, specific apoptosis and cell cycle analysis of all cell lines. (a) Growth curves over 120 h for all cell lines used in this study; physiological growth conditions. (b) Specific apoptosis rates shown for all cell lines over 120 h; physiological growth conditions; to calculate specific apoptosis rates the following equation was used: ((%Exp)À(%MEF/tTA))/(100À(%MEF/ tTA)) Â 100; MEF/tTA ¼ data obtained from untransfected cells grown under physiological conditions; Exp: data obtained from transfected cell lines. (c) MTT assay for all cell lines in this study; physiological growth conditions. (d) Growth curves over 120 h for all cell lines used in this study; growth conditions: serum deprivation. (e) Specific apoptosis rates shown for all cell lines over 96 h; growth conditions: serum deprivation; to calculate specific apoptosis rates the following equation was used: ((%Exp)À(%MEF/tTA))/ (100À(%MEF/tTA)) Â 100; MEF/tTA ¼ data obtained from untransfected cells grown under physiological conditions; Exp: data obtained from transfected cell lines under serum deprivation. (f) Cell cycle analysis for all cell lines used in this study; growth conditions: serum deprivation.

We next investigated all cell lines in serum-free media tTAHder4 clone that displayed apoptotic rates between (DMEM/1% Pen-Strep) to investigate their capability 2 and 43% over the observation period. By contrast, co- to cope with stress situations. All transgenes were transfected MEF/tTA cells displayed the lowest level of induced again for 5 days in the absence of doxycycline apoptosis in the range between 2 and 21%. As and then cultured in serum-free media. All subsequent summarized in Figure 2f, untransfected MEF/tTA cells experiments were repeated three times in triplicate. As mostly arrested in G0 (86–93% over 24–96 h). The shown in Figure 2d–f, cell growth, specific apoptosis MEF/tTAHder11 clone displayed arrested cells in the rates, and cell cycle analysis were performed over a time range of 54–77% (24–96 h). The MEF/tTAHder4 clone period of 24–96 h. Co-transfected cells displayed the displayed an even lower G0-rate between 52 and 68% lowest decrease of cell numbers under these conditions during the observation period. Co-transfected cells had (B28% reduction after 96 h). However, even the single- a stable G0-rate between 77 and 81%, but still exhibited transfected cells had only a reduction of 35–39% and 23–19% of cycling cells. differed significantly from control cells that displayed a 70% reduction of cell numbers (Figure 2d). As Focus formation experiments summarized in Figure 2e, the specific apoptosis rate All cell lines were seeded in small Petri dishes (1 Â 104 of control MEF/tTA cells was between 2 and 33% cells) and grown in the absence or presence of doxycy- (24–96 h), whereas the MEF/tTAHder11 clone showed a line for 21 days. All cells were formalin-fixed and stained lower apoptotic rate in the range between 2 and 23%. with methylene blue solution. As shown in Figure 3, in Highest rates were observed again for the MEF/ the presence of doxycycline (absence of transgene

Oncogene t(4;11) pathobiology A Gaussmann et al 3355

Figure 3 Focus formation experiments. Whole plate and microscopic pictures ( Â 100) taken from 60 mm Petri dishes after formaldehyde fixation and methylene blue staining. Panels in first and third lane: cell lines grown for 21 days in the presence of 10 mg/ ml doxycycline. Panels in second and fourth lane: cell lines grown for 21 days in the absence of doxycycline. Focus formation was observed only for AF4 . MLL- and co-transfected cell lines in the absence of doxycycline. expression), all cells were growing until confluency was reached. Without doxycycline in the media (all trans- genes expressed) loss-of-contact inhibition was observed for MEF/tTAHder4 and MEF/tTAHder4/11 cells, whereas MEF/tTAHder11 and untransfected cells did not show such a growth-transformed phenotype. This clearly demonstrates that AF4 . MLL- and co-trans- fected cells have the capability of overcoming contact inhibition, and thus AF4 . MLL has the dominant property to growth transform even in the presence of MLL . AF4. All experiments have been repeated at least for 3 times in triplicate and exemplary pictures are Figure 4 Q-PCR analysis of cell cycle and pro-/antiapoptotic shown. genes. Q-PCR experiments were performed in triplicate for cell cycle and apoptosis-related genes. For all measurements, triplicates of GAPDH were used to normalize the gene expression data. Quantitative-polymerase chain reaction analyses of (a) DDC(t) data of p16, p18, and p27. Significant changes were CDKIs, pro- and antiapoptotic genes observed for p16 in MLL . AF4-(der11)-transfected cells. p21 and To investigate the molecular effects on cell cycle and p27 were significantly lower in co-(der4/11)-transfected cells. apoptosis, several genes involved in these pathways were (b) DDC(t) data of Bim, Bid, Bax, Bak, Bcl2 and BclXL. Bim, Bid and Bcl2 were significantly lower in co-(der4/11)-transfected cells. analysed by quantitative-polymerase chain reaction Bax is relatively higher in AF4 . MLL-(der4)-transfected cells. (PCR) experiments. As outlined above, all analyses were performed after 5 days induction of transgenes. Again, all experiments were repeated three times in triplicate. As shown in Figure 4a, Q-PCR experiments of no significant differences, whereas the co-transfected several CDKIs (p16, p18, p21, p27) revealed that MEF/ MEF/tTAHder4/11 cells displayed significantly lower tTAHder11 cells displayed a 2.5-fold difference of transcription rates of p21CIP1 (0.2270.009) and p27KIP1 p16INK4A (2.5370.19). MEF/tTAHder4 cells displayed (0.2970.023), respectively.

Oncogene t(4;11) pathobiology A Gaussmann et al 3356 Surprisingly, all analysed pro- and antiapoptotic extracellular matrix/structure organization and biogen- genes displayed lower transcription rates in the trans- esis, and . The MEF/tTAHder4 clone fected cell lines (25–95% decrease) when normalized to exhibits a total of 660 differentially expressed genes (2- glyceraldehyde-3-phosphate dehydrogenase (GAPDH) to 398-fold differences in gene expression; 552 genes (the GAPDH mRNA levels were equal for all trans- were upregulated and 108 genes were downregulated). fected and untransfected cells as judged by the DC(t) The identified genes belong to the following ontogenetic values). In MEF/tTAHder4 cells, a relative activation of groups: cell adhesion, cell communication, defense Bax was observed when compared to the other response, sulfur metabolism and homophilic cell adhe- transfected cell lines. The observed reduction of sion. The co-transfected MEF/tTAHder4/11 clone dis- apoptosis in the co-transfected MEF/tTAHder4/11 cell played 800 differentially expressed genes (2- to 150-fold line may presumably be explained by the very low differences in gene expression; 677 genes were upregu- expression of Bim (0.0570.003) and Bid (0.0670.002), lated and 123 genes were downregulated). The identified both activators of Bax and Bak(Figure 4b). However, genes belong to the following ontogenetic groups: cell these data did not give a satisfactory explanation for the adhesion, cell communication, organismal physiological observed effects. processes, organ morphogenesis, organ development, immune response, defense response, cell–cell adhesion, Gene expression profiling experiments using single- and odontogenesis and skeletal development. co-transfected cells As shown in Figure 5a, the data sets displayed All cell clones were grown under optimal conditions intersections of overlapping gene expression patterns without doxycycline for 5 days and total RNA was between all three clones. Of interest, about 60% of all isolated. All RNA preparations were independently genes identified in the MEF/tTAHder4 clone are shared repeated three times and mixed to eliminate biological with the co-transfected MEF/tTAHder4/11 clone, fluctuations. The RNA pools were tested for endogen- whereas in MEF/tTAHder11 cells only a very few genes ous house keeping genes and optimal RNA quality. are shared with the other two clones. Moreover, all three Affymetrix chip experiments were performed twice by clones displayed individual gene signatures that dis- using Mouse Genome 430 2.0 arrays. Subsequent criminated the clones from each other (MLL . AF4: 337 hybridization experiments showed a total expression of genes; AF4 . MLL: 234 genes; co-transfected cells: 359 about 48% for all spotted genes in both experiments. All genes). The gene signatures A–G are listed in Supple- transfected cell lines were compared to the untransfected mentary Tables A–G respectively. MEF/tTA cell line. As shown in Figure 5a, a total of 562 Interestingly, the MEF/tTAHder11 cells displayed the differentially expressed genes were observed in the overexpression of only two Hox genes: HoxA7 was MEF/tTAHder11 cell clone (2- to 146-fold differences 3.3-fold overexpressed whereas HoxA10 was 6.1-fold in gene expression; 399 genes were upregulated and 163 overexpressed. No evidence for HoxA9 or Meis-1 genes were downregulated). The identified genes belong overexpression was found when compared to untrans- to the following ontogenetic groups: cell adhesion, cell fected control cells. Surprisingly, the HoxA7- and communication, organ development, organ - HoxA10-specific signature disappeared in the co-trans- esis, response to external stimuli, defense response, fected MEF/tTAHder4/11 cells. However, co-transfected

Figure 5 Results of gene expression profiling experiments. (a) Gene expression profiling data for all three transfected cells lines. In order to compare the different signatures by a database program, all duplicates and unnotated genes were eliminated. For MLL . AF4- transfected cells, 562 genes were identified; for AF4 . MLL-transfected cells, 660 genes were identified; for co-transfected cells 800 genes were identified. These genes were grouped according to their intersections with each other, resulting in gene signatures A–G. AF4 . MLL- and co-transfected cells share about 38% of their deregulated genes (252 out of 660), whereas MLL . AF4-transfected cells share about 6% with AF4 . MLL-transfected cells and about 9% with co-transfected cells. One hundred and thirty-eight genes are shared in all three transfected cell lines. (b) Each gene signature (A–G) was compared to data obtained from gene expression profiling of Mll k.o. cells (1) and the MllDSET mutant strain (2). In addition, primary target genes of MLL (3) and Nanog (4) were compared; the precise numbers and percentages of identified genes are listed.

Oncogene t(4;11) pathobiology A Gaussmann et al 3357 cells seem to overexpress the homeobox gene Nanog To gain confidence that there is biological activity of 16-fold. Nanog was identified as the key component Nanog in the co-transfected cell line, we first compared to maintain the pluripotency of the embryonic stem, the MEF/tTAHder4/11 gene expression profile with an together with Oct4 and Sox2 (Boyer et al., 2005; available Nanog data set obtained by another ChIP-on- Loh et al., 2006). Nanog/Oct4 expression confers self- CHIP experiment (1.409 genes; Boyer et al., 2005). renewal in a leukemia inhibitory factor (LIF)/STAT3- There is a median overlap of 6% with Nanog target independent manner and blocks specifically all differ- genes, indicating that established gene sets A–G entiation processes (Mitsui et al., 2003). represent a tiny fraction of primary Nanog target genes Next, we compared the data sets of all three lines with (Figure 5b). gene expression profiling data sets derived from the A total of 30, 46 and 55 Nanog target genes were analysis of Mll k.o. cells (Yu et al., 1995) and a recently identified in MLL . AF4-, AF4 . MLL- and co-trans- published DSET mutant of Mll (Terranova et al., 2006). fected cells, respectively. The highest percentages of The latter MLL mutant has viable offspring but does Nanog target genes were identified for gene sets C, F not exhibit the histone H3K4 methyltransferase activity. and G that define target genes idiosyncratic for Mll k.o. cells displayed 1.991 deregulated genes (2- to AF4 . MLL-, co-transfected cells and genes shared by 781-fold differences in gene expression; 1.171 genes were AF4 . MLL- and co-transfected genes. This may indicate upregulated and 820 genes were downregulated), that the presence of AF4 . MLL or of both MLL fusion whereas Mll DSET cells revealed 1.560 deregulated proteins is presumably key to activating the Nanog gene genes (2- to 333-fold differences in gene expression; 643 or one of its retrogenes. genes were upregulated and 917 genes were down- All data sets obtained from the three cell lines were regulated). The co-transfected MEF/tTAHder4/11 clone selectively analysed for the presence of genes involved in shares 254 target genes with Mll k.o. cells and 204 target transcription, chromatin modification, signaling, cell genes with Mll DSET cells (overlap of 31 and 26%). The cycle, apoptosis and DNA repair. As summarized in same was true for MEF/tTAHder4 that shared 225 and Table 1, we first analysed deregulated transcription 170 target genes with both MLL mutant strains (overlap factors. A small fraction is shared in all three cell lines of 34 and 26%; Figure 5b). Single-transfected cells (underlined). Each cell line displays a series of (mostly expressing the MLL . AF4 transgene shared 185 and 156 upregulated) transcription factors. Forty transcription genes with Mll k.o. and Mll DSET cells (overlap of 33 factors were identified in co-transfected cells (33 and 28%; Figure 5b). Thus, all transfected cells shared a upregulated, 7 downregulated), whereas MLL . AF4- significant portion of their deregulated genes (26–34%) and AF4 . MLL-transfected cells displayed 33 and 36 with genes deregulated by knocking out (functions of) deregulated transcription factors, respectively (MLL . AF4: the MLL gene. This may indicate that the presence of 26 upregulated, 7 downregulated; AF4 . MLL: 30 upregu- t(4;11) fusion proteins in the single- and co-transfected lated, 6 downregulated). Seven transcription factors were clone mimic, in part, an Mll k.o. phenotype, presumably deregulated in all three lines (Id4, Egr3, Asb5, Fos, Hdac9, due to a competitive situation between the MLL Hey1 and Trib3). multiprotein complex and the protein complexes form- Each cell line also displayed transcription factors ing upon the two reciprocal MLL fusion proteins. that are controlled by the core Nanog network We next compared gene sets A–G with primary MLL (MLL . AF4: 1; AF4 . MLL: 4; co-transfected: 5; marked target genes that were recently identified by a ChIP-on- in bold or bold italics). However, only AF4 . MLL- CHIP experiment using antibodies against MLL protein and co-transfected cells exhibit Nanog targets that (2.420 genes; Guenther et al., 2005). There is a median are transcriptionally activated by the core Nanog overlap of 9% with these MLL target genes, indicating network( Jarid2, Nanog, Stat3, Zfp36l1 and Zic3), that the established gene sets A–G represent a small while the only core Nanog target gene identified fraction of primary MLL target genes (Figure 5b), in MLL . AF4-transfected cells, the segment polarity whereas non-overlapping genes observed in the gene gene , is normally repressed by the core expression profiles presumably reflect secondary target Nanog network. The same was true for co-transfected genes. cells that express the potent oncogene Fli1, normally One of the top-scoring genes identified in the co- repressed in Nanog-expressing cells. The expression transfected clone was Nanog. This gene is upregulated of these target genes may indicate that they are about 16-fold, whereas any of the single-transfected cells either not repressed by the Nanog system or did not show any differences in Nanog transcription. We arbitrarily activated by the presence of the t(4;11) fusion have to mention that there are 11 pseudogenes of Nanog proteins. that differ by distinct point , small deletions Table 2 summarizes the overlapping transcription, and frameshifts. However, two of them are so-called chromatin modification, signaling, cell cycle, apoptosis retrogenes (Nanog P8, P10) that encode proteins with a and DNA repair genes of the established cell lines versus 99% identity to the bona fide Nanog protein (Booth and target genes identified by gene expression profiling Holland, 2004). The Nanog P8 retrogene is biological, experiments using the Mll k.o. strain, the Mll DSET functional and able to substitute for the functions of the mutant strain, or primary target genes identified by bona fide Nanog protein (Zhang et al., 2006). All the ChIP-on-CHIP experiments performed by others using other pseudogenes encode C-terminal truncated versions anti-MLL or anti-Nanog antibodies (Boyer et al., 2005; of the Nanog protein. Guenther et al., 2005).

Oncogene t(4;11) pathobiology A Gaussmann et al 3358 Table 1 Transcription factors identified in transfected cell lines Table 2 Overlapping genes between transfected cells lines, MLL mutant strains and primary MLL or core Nanog target genes der11 Logchange der4 Logchange der4/11 Logchange TxF Chrom Sig CC Apop DNA Cri2 +6.00 Hdac9 +4.94 Lhx9 +5.78 Runx3 +4.02 Zic3 +3.43 Phf14 +4.59 MLL . AF4 33 3 22 17 13 1 En1 +3.79 Lhx9 +3.25 Nanog +4.02 MLL k.o. 12 1 1 5 6 0 Id4 +3.45 Fos +3.04 Id4 +3.43 MLLDSET 10 0 1 4 5 0 Jarid1c +3.14 Bhlhb5 +2.92 Cebpa +3.09 MLL targets 5 1 0 3 1 0 HoxA10 +2.62 Sox7 +2.41 Fos +2.84 Core Nanog targets 1 0 0 0 0 0 Pitx2 +2.48 Tcf7 +2.23 Fli1 +2.64 Egr3 +2.26 Zfp36l1 +2.01 Myst3 +2.49 AF4 . MLL 36 2 31 17 22 3 Asb5 +1.78 Asb5 +1.95 Foxq1 +2.38 MLL k.o. 10 1 16 8 13 1 Huwe +1.75 Id4 +1.67 Foxj1 +2.30 MLLDSET 6 0 7 8 3 1 HoxA7 +1.73 Btg2 +1.63 Hdac9 +2.15 MLL targets 6 1 3 4 1 2 Myocd +1.70 Tle2 +1.57 Zfp36l1 +1.92 Core Nanog targets 4 0 1 0 0 0 Fos +1.63 Jarid2 +1.55 Egr3 +1.84 Hdac9 +1.46 Nfkbiz +1.39 Nsd1 +1.62 MLL . AF4/AF4 . MLL 40 3 34 19 19 2 Nr3c1 +1.39 klfl2 +1.37 Bhlhb5 +1.55 MLL k.o. 13 1 12 8 11 1 Ebf1 +1.31 Rfx3 +1.35 Nfyg +1.53 MLLDSET 10 0 9 8 5 1 Gata6 +1.31 Egr3 +1.26 Btg2 +1.48 MLL targets 10 1 4 4 3 1 Zfp198 +1.24 Runx3 +1.26 Asb5 +1.34 Core Nanog targets 5 1 0 0 0 0 Ssbp2 +1.18 Zfp398 +1.26 Nfkbiz +1.34 Wasl +1.17 Nfkbie +1.23 Vdr +1.30 Number of genes identified in the transfected cell lines were selected Hdac2 +1.15 Cepbd +1.20 Zfp398 +1.24 according to the following ontological groups: transcription factors Id2 +1.14 Glis3 +1.17 Rpo2tc1 +1.22 (TxF), chromatin modifying proteins (Chrom), cell signaling (Sig), cell Zfp422 +1.11 Hey1 +1.16 Cepbd +1.21 cycle (CC), apoptosis-related genes (Apop) and DNA repair (DNA), Hey1 +1.09 Tcf4 +1.16 Hey1 +1.20 respectively. Below: overlap with genes identified either by gene Egr1 +1.04 Stat3 +1.16 Ssbp2 +1.20 expression profiling using two Mll mutant strains or the available data Brd8 +1.03 Nsd1 +1.15 Egr2 +1.19 obtained by ChIP-on-CHIP experiments by using antibodies directed HoxD13 À1.04 E2f6 +1.15 Nrarp +1.16 against MLL or Nanog; for Nanog only the 353 identified genes of the Hira À1.14 Per1 +1.13 Eya3 +1.14 core Nanog networkwere used for comparison. Neo1 À1.14 Ches1 +1.11 Zbtb1 +1.11 À1.22 Ppp1r13b +1.04 Nfib +1.11 Atf3 À1.35 Atf5 À1.05 Jarid2 +1.01 Trib3 À1.60 Twist2 À1.08 Tle2 +1.01 DNA binding protein that is involved in the transcrip- Esr1 À2.66 Tle4 À1.22 Nfx1 +1.01 tional activation of the Nanog gene, but is presumably Neo1 À1.20 Neo1 À1.12 Trib3 À2.13 Twist2 À1.16 dispensable if Nanog (or a retrogene thereof) is already Dach2 À4.70 Atf5 À1.20 transcriptionally active. Several target genes of the core Ing5 À1.36 Nanog networkwere also investigated ( Stat3, Myst3, Trib3 À2.02 REST, ZFP36L1, Meis1, Fli1, Pax6, ESX1L). All tested Etv1 À2.47 Dach2 À4.44 downstream target genes of the core Nanog network were found to be transcriptionally active in t(4;11) patient cells. This indicated that the observed transcripts Transcription factors identified in MLL . AF4-, AF4 . MLL- and co- of Nanog and Oct4 in t(4;11) patient cells were coding transfected cells. Underlined: genes deregulated in all three lines. Bold for functional proteins, sufficient to initiate and main- italics: genes that are repressed by the core Nanog regulatory network. tain transcription of genes known to belong to the stem Bold: genes that are transcriptionally activated by the core Nanog regulatory network. cell compartment. In conclusion, the observed transcrip- tion of Nanog (or of functional Nanog retrogenes) in conjunction with Oct4, resulting in a stem-cell-like activity, is presumably an important and novel feature Confirmation of Nanog expression in biopsy material of of t(4;11) leukemia. t(4;11) patients To verify ectopic Nanog expression in t(4;11) leukemic cells, we next analysed Nanog transcription in biopsy material of adult and pediatric t(4;11) patients. We have Discussion chosen only t(4;11) patients that were positively analysed for the presence of both reciprocal MLL We established a conditional ‘in vitro t(4;11) model fusion alleles (MLL . AF4 þ /AF4 . MLL þ patients). As system’ to study certain aspects of t(4;11) biology. We shown in Figure 6, all 14 t(4;11) patients (10 adult, 4 could show that both t(4;11) fusion proteins exhibit pediatric patients) displayed transcription of the Nanog oncogenic properties. The MLL . AF4 fusion protein gene (or of full-length retrogenes such as Nanog P1 or increases resistance against apoptosis (which is domi- Nanog P8). Moreover, t(4;11) patient cells showed nant over AF4 . MLL), whereas the AF4 . MLL fusion active transcription of Oct4 and Bmi1, two genes protein is able to growth transform cells (which is selectively expressed in stem cells. However, no tran- dominant over MLL . AF4). Single-transfected t(4;11) scriptional activity was found for Sox2, the third fusion proteins were able to affect cell growth and cell component identified in the core Nanog network(Boyer cycling. However, the co-transfected cell line displayed et al., 2005; Loh et al., 2006). Sox2 is a minor-groove the highest growth rate, maintained cell cycling in the

Oncogene t(4;11) pathobiology A Gaussmann et al 3359 described (Caslini et al., 2004). The upregulation of p16 in MLL . AF4-transfected cells could explain the observed growth inhibition, which is also in line with published data (Caslini et al., 2004). It has already been shown that the MLL . AF4 fusion protein directly binds to the regions of two cell cycle inhibitors, p18 and p27, and regulates their gene expression (Xia et al., 2005). Recently, functional data were obtained from murine embryonic fibroblasts derived from Taspase1 knockout mice (Takeda et al., 2006). Taspase1-deficient cells and MLL . AF4-transfected cells are both growth inhibited and display overexpression of p16. Thus, cells expressing the MLL . AF4 fusion protein and cells that produce only uncleaved MLL protein behave very similarly. The AF4 . MLL-single-transfected cell line mimics the situation of t(4;11) SEM cells after downregulation of the MLL . AF4 fusion protein by using an siRNA- mediated approach (Thomas et al., 2005). Transient downregulation of the MLL . AF4 fusion protein resulted in a strong increase of apoptosis, a typical feature identified for the single-transfected MEF/tTAH der4 cells. Assuming that Nanog (or Nanog-like activity) – identified only in the co-transfected cells – is required for clonogenicity, it would explain the observed Figure 6 Exploring the core Nanog regulatory networkin t(4;11) reduction of the replating efficiency of these siRNA- leukemia cells. Upper panel: cDNA obtained from t(4;11) patients treated SEM cells (Thomas et al., 2005). One of the were used to identify Nanog transcripts in primary biopsy material biological features of MLL . AF4-bearing cells is their from 10 adult and four pediatric t(4;11) leukemia patients (all exhibit both reciprocal fusion alleles). The oligonucleotides reduced sensibility against apoptosis. This feature is also amplified a 419 bp Nanog amplimer in all patients. Additional reflected by two recently established transgenic mouse bands: of Nanog 4. Other bands (below model systems using the MLL . AF4 transgene (Chen and above) are derived from expressed Nanog retrogenes. Lower et al., 2006; Metzler et al., 2006). In both mouse models, panels: RT–PCR experiments using oligonucleotides specific for a B-cell lymphoma phenotype was observed due to the Pou5f1/Oct4 (440 bp), Sox2 (398 bp), Bmi1 (451 bp), Stat3 (327 bp), Myst3 (590 bp), ZFP36L1 (449 bp), REST (447 bp), Meis1 (503 bp), expression of the transgene. However, the disease Fli1 (441 bp), Pax6 (399 bp), ESX1L (380 bp) and b-actin control phenotype had a low penetrance and an extremely long (347 bp). In all patients, consistent expression of all genes could be latency (50% lymphomas after a median time of demonstrated. Left: blackboxes indicate that these particular genes 520–540 days). Frequently, lymphomas derive from should be actively transcribed in the presence of functional Nanog/ Oct4; shaded boxes indicate that these particular genes should be germinal center B cells which acquired a blockin one of transcriptionally repressed in the presence of functional Nanog/ their apoptotic pathways and therefore arbitrarily Oct4; white boxes: genes not controlled by the Nanog/Oct4 survived apoptotic triggers due to a selective advantage network. (e.g. a chromosomal translocation). Assuming that overexpression of MLL . AF4 mimics in part the dominant phenotype of overexpressed antiapoptotic genes (Strasser et al., 1990), the development of a absence of growth factors and showed the highest lymphoma disease phenotype is not unexpected. resistance against apoptotic triggers. Moreover, co- In addition, cells of both transgenic mouse models transfected cells exhibited a loss-of-contact inhibition, showed neither a differentiation arrest nor the accumu- another typical feature of transformed cells. Thus, lation of immature cells, arguing for a missing second MEF/tTAHder4/11 cells combined properties observed hit. In case of t(4;11) leukemia the missing second hit for both single-transfected cell lines, and also gained probably derives from the reciprocal AF4 . MLL fusion additional properties that were not observed in the allele. Thus, the t(4;11) chromosomal translocation single-transfected cells (see below). creates two independent hits, which cause enhanced The presented data are in line with all yet published and uncontrolled and resistance data. Co-transfected MEF/tTAHder4/11 cells displayed against apoptosis at the same time. It has already been a higher resistance against apoptosis, comparable with shown that Taspase1-mediated proteolysis of the data obtained in another study, demonstrating that AF4 . MLL fusion protein results in the accumulation t(4;11) cell lines have the highest resistance against drug- of the AF4 . MLL fusion protein and thereby provides induced apoptosis when compared with cells expressing important oncogenic functions (Bursen et al., 2004). The other leukemogenic gene fusions (Kersey et al., 1998). data obtained from the focus formation experiments A lower expression of the antiapoptotic Bcl-2 protein indicate that this biological property of the AF4 . MLL in MLL . AF4-transfected U937 cells has also been fusion protein even dominates over the MLL . AF4

Oncogene t(4;11) pathobiology A Gaussmann et al 3360 fusion protein. In addition, the two reciprocal MLL investigated to unravel the extent of deregulated target fusion proteins cooperate in a synergistic manner genes. because co-transfected cells display enhanced properties Another interesting observation is the overexpression (growth rate, resistance against apoptosis) or features of HoxA7 and HoxA10, but not of HoxA9 in not observed before in the two single-transfected cell MLL . AF4-transfected cells. HoxA9 is the major target lines (e.g. the downregulation Bim, Bid and Bcl2; gene that is transcriptionally activated by many different downregulation of p21 and p27; transcriptional activa- MLL fusion proteins such as MLL . ENL, MLL . AF6, tion of Nanog, Oct4, Bmi1 and several other ). MLL . AF10, MLL . ELL and MLL . CBP (Ayton and One important finding is the transcriptional upregula- Cleary, 2003). However, overexpression of HoxA7 and tion of Nanog (or Nanog-like transcripts) in co- HoxA9 is always correlated with AML. We analysed transfected cells. Ectopic expression of Nanog leads to MLL fusion proteins associated with an ALL disease dedifferentiation and blocks further developmental phenotype. In a recent study comparing gene expression processes (Taranger et al., 2005). Malignant transforma- profiles of acute lymphoblastic leukemia patients, a tion of teratocarcinomas and germline tumors are HoxA7/HoxA10 signature was identified in MLL-rear- caused by ectopic expression of Nanog (Almstrup ranged ALL when compared with other B-lineage ALL et al., 2004; Hoei-Hansen et al., 2005; Skotheim et al., or AML (Andersson et al., 2005; Supplementary Tables 2005). Overexpression of the full-length retrogene 21–23). Noteworthy, co-transfected cells lost this Hox Nanog P8 has been identified in a large variety of signature, but showed an upregulation of the homeobox different solid tumors or cancer cell lines (Zhang et al., gene Nanog. 2006). This finding also sheds light on the ongoing The latter finding points to a mechanism where the discussion about cancer stem cells (Jordan et al., 2006). presence of the MLL . AF4 fusion protein or both Systematic analysis of more cancer types and liquid reciprocal t(4;11) fusion proteins may be of critical tumors will presumably make clear that an arbitrary importance. If both reciprocal MLL fusion proteins are expression of Nanog or certain Nanog retrogenes might expressed, a ‘stem cell-like program’ will be induced, be a universal mechanism for cancer cells to generate whereas the overexpression of the MLL . AF4 fusion pluriopotency. protein alone is able to initiate the ‘Hox program’. This Nanog – in combination with Oct4 and Sox2 – interpretation solves the controversial situation found initiates a genetic program that allows embryonic stem for 20% of t(4;11) leukemia patients, where only the cells to self-renew and to blockdifferentiation processes presence of the MLL . AF4 fusion allele was successfully (Ivanova et al., 2006). Recently, this ‘core Nanog verified (Downing et al., 1994; Janssen et al., 1994; network’ has been unraveled (Boyer et al., 2005). The Reichel et al., 2001). In both genetic situations self- core Nanog networkcontrols about 350 genes, of which renewal would be granted, either by a Nanog-dependent about 100 are transcription factors. As the activated ‘stem cell program’ or by HoxA genes in conjunction genes of the core Nanog networkconfer stemness, it is with Meis1. As overexpressed AF4 . MLL fusion protein plausible that t(4;11) cells exhibiting Nanog- and Oct4- confers sensitivity against apoptosis, downregulation or like activities may induce self-renewal in a similar of this fusion gene might even be a selective fashion. advantage for leukemic cells. To verify these findings for t(4,11) leukemia, we In conclusion, we were able to demonstrate for the analysed the transcription of several Nanog/Oct4 target first time that the pathological disease mechanism of genes in biopsy material of t(4;11) patients. As shown in t(4;11) leukemia depends on the combined biological Figure 6, adult and infant t(4;11) patient cells actively properties of both reciprocal MLL fusion proteins. To transcribe Nanog, Oct4 and Bmi1, respectively. Several verify this finding, retroviral transduction experiments target genes of the core Nanog networklike Stat3, Myst3, of murine hematopoietic stem cells are currently under- REST and ZFP36L1 are consistently transcribed. This way to unravel the genetic prerequisites for t(4;11) indicates that Nanog-andOct4-derived activities are leukemia. Only these in vivo experiments will finally present in the analysed t(4;11) leukemia cells. answer the interesting question whether one or both of By contrast, Meis1 and Fli1, two genes that are the t(4;11) fusion proteins will be sufficient and normally repressed by PRC2 (Lee et al., 2006), were necessary to establish leukemia in an animal model actively transcribed in all investigated t(4;11) biopsy system. probes. This may indicate that specific Nanog/Oct4 target genes may be deregulated in the presence of the t(4;11) fusion proteins. As a matter of fact, 19 out of 57 Materials and methods repressed Nanog/Oct4 target genes are also primary targets of the MLL protein (ATBF1, DLX5, EN1, FLI1, Patient material FOXA2, FOXD3, GBX2, GLI3, HOXB3, HOXC4, Total RNA was isolated from and/or peripheral INSM1, IPF1, LHX2, MEIS1, NR4A2, NR6A1, RORB, blood samples of t(4;11) patients and cDNA syntheses were performed. Informed consent was obtained from all patients or SOX5, TBX5; Guenther et al., 2005). Thus, the presence patients’ parents/legal guardians and control individuals. of the t(4;11) fusion proteins may counteract the transcriptional repression of Fli1 and Meis1. The Cloning of cDNA expression constructs transcriptional properties of all transcription factors The cDNA cassettes coding for MLL . AF4 (der11) and regulated by the core Nanog networkare currently AF4 . MLL (der4) have been described (Bursen et al., 2004).

Oncogene t(4;11) pathobiology A Gaussmann et al 3361 Using the flanking SfiI-sites, both constructs were cloned after Quantitative RT–PCR experiments were performed using insertion of a compatible SfiI-linker into the multiple cloning the Mini-Opticon-Cycler (Biorad, Munich, Germany) and site of pTRE2-puro and pTRE2-hygro (BD Clontech, Heidel- Sybrgreen Mix (Biorad) according to the manufacturer’s berg Germany), respectively. Both vectors contain a tetracyclin instructions. All experiments were performed in triplicates. response element designed for the conditional Tet-off gene GAPDH served as an internal standard to normalize the gene expression system (BD Clontech) according to the manufac- expression levels. The following primers were used: Bak . 3 turer’s instructions. The MLL . AF4 cDNA cassette was (50-TGACAAGTGACGGTGGTCTC-30), Bak . 5(50-ACCAC cloned into the pTRE2-hygro the AF4-MLL cDNA cassette GCTGGTAGACGTACA-30); Bax . 3(50-CCACCAGCTCTG into the pTRE2-puro vector. AACAGAT-30), Bax . 5(50-ACGGAGGAAGTCCAGTGTC- 30); Bim . 3(50-TACAGACAGAACCGCAAGAC-30), Bim . 5 Cell culture and stable of MEF/tTA cells (50-CTCCATACCAGACGGAAGAT-30); Bid . 3(50-ATCA For generating stably transfected cell lines expressing CAGACCTGCTGGTGTT-30), Bid . 50 (50-GAAGGCTGTCT MLL . AF4, AF4 . MLL or both fusion proteins, we used TCACCTCAT-30); Bcl2 . 3(50-CATTGGTACCTGCAGCT commercially available MEF/tTA cells (BD Clontech). MEF/ TCT-30), Bcl2 . 5(50-GCAGCTGACTGGACATCTCT-30); tTA cells were grown in DMEM/10% FCS supplemented with BclXL . 3(50-AGTGAAGCAAGCGCTGAGAG-30), BclXL . 1% Pen-Strep and 100 mg/ml G418. For stable transfection the 5(50-TCAGGAACCAGCGGTTGAAG-30); p16 . 3(50-GTCG FUGENE 6 Transfection Reagent (Roche, Mannheim, CAGGTTCTTGGTCACT-30), p16 . 5(50-TCGCACGATG Germany) was used, following the instructions of the TCTTGATGTC-30); p18 . 3(50-GAGAACTGCGCTGCAG manufacturer. To establish the MEF/tTAHder4 and MEF/ GTTA-30), p18 . 5(50-TAGAACCTGGCCAAGTCGAA-30); tTAHder 11 cell lines, selection was started 48 h after p21 . 3(50-AAGTGTGCCGTTGTCTCTTC-30), p21 . 5(50-AT transfection with 300 mg/ml hygromycin or 3 mg/ml puromycin, CTGCGCTTGGAGTGATAG-30); p27 . 3(50-GATACGAGT respectively. Using two antibiotics, a co-transfected cell line GGCAGGAGGT-30), p27 . 5(50-TTACGTCTGGCGTCGA containing both conditional cDNA cassettes was established AG-30); murine Nanog: mNanog . 3(50-GCCTCCAGCAGAT (MEF/tTAHder4/11). Single clones were established for each GCAAGAAC-3), mNanog . 5(50-GCAGAGAAGTTTTGC line and tested by reverse transcription (RT)–PCR experiments TGCAAC-30). for their ability to induce transgene expression. All transgenes were suppressed by adding 10 mg/ml doxycycline to the media. Focus formation experiments For each experiment described below, the expression of all For focus formation assays, 1 Â 104 cells were seeded on 60 mm transgenes was induced by culturing the stably transfected cell dishes (Corning). Cells were grown in DMEM supplemented with lines for 5 days in the absence of any doxycycline. 5% FCS, 5% CS, the appropriate antibiotics and in the presence or absence of doxycycline. Culture medium was exchanged every RT–PCR experiments 3 days. After 14–21 days, cells were washed once with PBS and For RT–PCR experiments, RNA was isolated using the then fixed with 3 ml of a 2% formaldehyde solution for 2 min. RNeasy Kit (Qiagen, Hilden, Germany) and the RNase free Thedisheswererinsedwithwaterandfixedcellswerestained DNase Kit (Qiagen), according to the manufacturer’s recom- with 0.5 ml of a 1 mg/ml methylene blue solution for 15 min. mendations. One microgram RNA was reverse transcribed into cDNA using superscript reverse transcriptase (Invitrogen, MTT assay Karlsruhe, Germany) following the manufacturer’s instruc- Cells were plated in triplicate at 4 Â 103 cells/well in a 96-well tions. cDNA from patient samples was obtained from t(4;11) plate. Over a period of 96 h, cell proliferation was measured patients and used for analyses. For RT–PCR experiments we every 24 h by adding 20 ml MTT (MERCK, Darmstadt, used the following oligonucleotides: MLL . AF4: MLL6 . 3(50- 0 . 0 Germany) in a concentration of 10 mg/ml in each well and CCTGTCACTAGAAACAAGGC-3 ), AF4 52 (5 -GAAT incubating for 3 h at 371C. The supernatant was removed and CTCTTCAACACAATGGAC-30); AF4 . MLL: AF4 . 324b 0 0 . 0 the precipitate was solubilized with 100 ml DMSO (100%) per (5 -GACAGATTCGGCTCCAGAG-3 ), MLL12 5(5-GGA well and the absorption was measured at 450 nm using an AGGGCTCACAACAGAC-30); human Nanog: hNANOG . 3 0 0 . 0 ELISA reader (Pegasus Scientific, MD, USA). All experiments (5 -GCCTCCAGCAGATGCAAGAAC-3 ), hNANOG 5(5- were repeated five times. GCAGGAGAATTTGGCTGGAAC-30); OCT4 . 3(50-GAGC AATTTGCCAAGCTCCTGA-30), OCT4 . 5(50-GCATAGTC GCTGCTTGATCGC-30); SOX2 . 3(50-AGCGCTGCACAT Analysis of apoptosis and cell cycle by flow cytometry GAAGGAGC-30), SOX2 . 5(50-CCTGCTGCGAGTAGGAC For cell cycle analysis and apoptosis measurements, the FITC ATG-30); MEIS1 . 3(50-CTCCTCTGCACTCGCATCAG-30), BrdU Flow Kit (BD Pharmingen, Heidelberg, Germany) was MEIS1 . 5(50-CTGGTCAGTTAGATTTGCTGATC-30); BMI1 . used. Cells treated as indicated above were pulse-labeled with 3(50-GCAGCTCATCCTTCTGCTGATG-30), BMI1 . 5(50- 10 mM BrdU for 30 min, fixed, permeabilized and stained with GTTCTCCAGCATTTGTCAGTCC-30); FLI1 . 3(50-AGTA anti-BrdU-FITC and 7-AAD according to the manufacturer’s 4 TGACCACATGAATGGATCC-30), FLI1 . 5(50-CACTCAA instructions. At least 10 cells were analysed by flow cytometry TCGTGAGGATTGGTC-30); MYST3 . 3(50-GTTGAACTT (FACScalibur and CellQuest Software, Becton Dickinson, CAAGGAGACATTGC-30), MYST3 . 5(50-CCATCTGTAG Heidelberg, Germany). Cells with DNA content oG1 were TTTCTGTCTTGG-30); Zfp36l1 . 3(50-CAGCTCCGTCTTG scored as apoptotic. Cell cycle analysis was performed excluding TACTTGG-30), Zfp36l1 . 5(50-CCACCATCTTCGACTTG cells with DNA content oG1 and >G2. All experiments were AGC-30); Stat3 . 3(50-GTCATCCTGGAGATTCTCTACC- performed in triplicate and repeated three times. 30), Stat3 . 5(50-GAGATTGACCAGCAGTATAGC-30); Pax6 . 3(50-CAGGTATTACGAGACTGGCTCC-30), Pax6 . 5 Gene expression profiling experiments (50-CTGATGGAGTTGGTATTCTCTCC-30); ESX1L . 3(50- Total RNA was isolated from untransfected and stably GAGCTGAAGCAAGAGCAGGAGG-30), ESX1L . 5(50-GT transfected cell lines on 3 independent days, mixed and used for GGCATAGGTGCTATGGGTGG-30); REST . 3(50-GT hybridization of Mouse Genome 430 2.0. arrays. All experiments GAGCGAGTATCACTGGAGG-30), REST . 5(50-GATACT were replicated twice in order to identify up- and downregulated GTAGATTACACTTCTTGG-30). genes. Gene lists were created by using only significantly changed

Oncogene t(4;11) pathobiology A Gaussmann et al 3362 gene signatures (D>2-fold). All Affymetrix experiments were mouse. We are grateful to Richard Young for providing performed on the NGFN II Affymetrix working platform in access to his lab, owing to which his Chip-on-CHIP data Essen (L Klein-Hitpass). Comparison analysis was performed by were publically available. We thankT Burmeister (Berlin) using the FileMaker database program. who provided the cDNAs from adult t(4;11) patients, and Andre´ Schrauder (Kiel) who provided the cDNAs Acknowledgements from infant t(4;11) patients. This study was supported by grants MA 1876/8-1 from the DFG, N1KR-S12T13 from We thankDr MalekDjabali, INSERM Marseille, for the BMBF, and 2001.061.2 from the Wilhelm-Sander-Founda- providing and sharing yet unpublished data on the MLLDSET tion to RM.

References

Almstrup K, Hoei-Hansen CE, Wirkner U, Blake J, Schwager Hoei-Hansen CE, Almstrup K, Nielsen JE, BraskSonne S, C, Ansorge W et al. (2004). -like features Graem N, Skakkebaek NE et al. (2005). Stem cell of testicular in situ revealed by genome-wide gene pluripotency factor NANOG is expressed in human fetal expression profiling. Cancer Res 64: 4736–4743. gonocytes, testicular carcinoma in situ and germ cell Andersson A, Olofsson T, Lindgren D, Nilsson B, Ritz C, tumours. Histopathology 47: 48–56. Ede´ nPet al. (2005). Molecular signatures in childhood acute Ivanova N, Dobrin R, Lu R, Kotenko I, Levorse J, DeCoste C leukemia and their correlations to expression patterns in et al. (2006). Dissecting self-renewal in stem cells with RNA normal hematopoietic subpopulations. Proc Natl Acad Sci interference. Nature 442: 533–538. USA 102: 19069–19074. Janssen JW, Ludwig WD, Borkhardt A, Spadinger U, Rieder H, Ayton PM, Cleary ML. (2003). Transformation of myeloid Fonatsch C et al. (1994). Pre-pre-B acute lymphoblastic progenitors by MLL oncoproteins is dependent on Hoxa7 leukemia: high frequency of alternatively spliced ALL1-AF4 and Hoxa9. Genes Dev 17: 2298–2307. transcripts and absence of minimal residual disease during Booth HA, Holland PW. (2004). Eleven daughters of complete remission. Blood 84: 3835–3842. NANOG. Genomics 84: 229–238. Jordan CT, Guzman ML, Noble M. (2006). Cancer stem cells. Boyer LA, Lee TI, Cole MF, Johnstone SE, Levine SS, Zucker N Engl J Med 355: 1253–1261. JP et al. (2005). Core transcriptional regulatory circuitry in Kersey JH, Wang D, Oberto M. (1998). Resistance of t(4;11) human embryonic stem cells. Cell 122: 947–956. (MLL-AF4 fusion gene) leukemias to stress-induced cell Bursen A, Moritz S, Gaussmann A, Moritz S, Dingermann T, death: possible mechanism for extensive extramedullary MarschalekR. (2004). Interaction of AF4 wildtype and accumulation of cells and poor prognosis. Leukemia 12: AF4 . MLL fusion protein with SIAH proteins: indication 1561–1564. for t(4;11) pathobiology? Oncogene 23: 6237–6249. Lange B, Valtieri M, Santoli D, Caracciolo D, Mavilio F, Caslini C, Serna A, Rossi V, Introna M, Biondi A. (2004). Gemperlein I et al. (1987). Growth factor requirements of Modulation of cell cycle by graded expression of MLL-AF4 childhood acute leukemia: establishment of GM-CSF- fusion oncoprotein. Leukemia 18: 1064–1071. dependent cell lines. Blood 70: 192–199. Chambers I, Colby D, Robertson M, Nichols J, Lee S, Lee TI, Jenner RG, Boyer LA, Guenther MG, Levine SS, Tweedie S et al. (2003). Functional expression cloning of Kumar RM et al. (2006). Control of developmental Nanog, a pluripotency sustaining factor in embryonic stem regulators by polycomb in human embryonic stem cells. cells. Cell 113: 643–655. Cell 21: 301–313 (doi 10.1016/j.cell.2006.02.043). Chen W, Li Q, Hudson WA, Kumar A, Kirchhof N, Kersey JH. Loh YH, Wu Q, Chew JL, Vega VB, Zhang W, Chen X et al. (2006). A murine Mll-AF4 knockin model results in lymphoid (2006). The Oct4 and Nanog transcription networkregulates and myeloid deregulation and hematological malignancy. pluripotency in mouse embryonic stem cells. Nat Genet 5: Blood March 21 (e-pub ahead of print). March (e-pub ahead of print). Dordelmann M, Reiter A, Borkhardt A, Ludwig WD, Gotz N, Metzler M, Forster A, Pannell R, Arends MJ, Daser A, Viehmann S et al. (1999). Prednisone response is the Lobato MN et al. (2006). A conditional model of MLL-AF4 strongest predictor of treatment outcome in infant acute B-cell tumourigenesis using invertor technology. Oncogene lymphoblastic leukemia. Blood 94: 1209–1217. 25: 3093–3103. Downing JR, Head DR, Raimondi SC, Carroll AJ, Curcio- Mitsui K, Tokuzawa Y, Itoh H, Segawa K, Murakami M, Brint AM, Motroni TA et al. (1994). The der(11)-encoded Takahashi K et al. (2003). The homeoprotein Nanog is MLL/AF-4 fusion transcript is consistently detected in required for maintenance of pluripotency in mouse epiblast t(4;11)(q21;q23)-containing acute lymphoblastic leukemia. and ES cells. Cell 113: 631–642. Blood 83: 330–335. Mitterbauer-Hohendanner G, Mannhalter C. (2004). The Felix CA, Hosler MR, WinickNJ, Masterson M, Wilson AE, biological and clinical significance of MLL abnormalities Lange BJ. (1995). ALL-1 gene rearrangements in DNA in haematological malignancies. Eur J Clin Invest 34: 12–24. topoisomerase II inhibitor-related leukemia in children. Pui CH, Behm FG, Downing JR, HancockML, Shurtleff SA, Blood 85: 3250–3256. Ribeiro RC et al. (1994). 11q23/MLL rearrangement confers Greil J, Gramatzki M, Burger R, Marschalek R, Peltner M, a poor prognosis in infants with acute lymphoblastic Trautmann U et al. (1994). The acute lymphoblastic leukemia. J Clin Oncol 12: 909–915. leukaemia cell line SEM with t(4;11) chromosomal rearran- Reichel M, Gillert E, Angermu¨ ller S, Hensel JP, Heidel F, gement is biphenotypic and responsive to -7. Br J Lode M et al. (2001). Biased distribution of chromosomal Haematol 86: 275–283. breakpoints involving the MLL gene in infants versus Guenther MG, Jenner RG, Chevalier B, Nakamura T, Croce children and adults with t(4;11) ALL. Oncogene 20: 2900– CM, Canaani E et al. (2005). Global and Hox-specific roles 2907. for the MLL1 methyltransferase. Proc Natl Acad Sci USA Skotheim RI, Lind GE, Monni O, Nesland JM, Abeler VM, 102: 8603–8608. Fossa SD et al. (2005). Differentiation of human embryonal

Oncogene t(4;11) pathobiology A Gaussmann et al 3363 in vitro and in vivo reveals expression profiles reprogramming by extracts of carcinoma and embryonic relevant to normal development. Cancer Res 65: 5588–5598. stem cells. Mol Biol Cell 16: 5719–5735. So ECW, Wong P, Lin M, Cleary ML. (2004). Disease models Terranova R, Agherbi H, Boned A, Meresse S, Djabali M. and transformation mechanisms mediated by MLL-AF4 (2006). Histone and DNA defects at Hox genes family oncoproteins in human leukemia. ASH Abst 467. in mice expressing a SET domain-truncated form of Mll. Stong RC, Korsmeyer SJ, Parkin JL, Arthur DC, Kersey JH. Proc Natl Acad Sci USA 103: 6629–6634. (1985). Human acute leukemia cell line with the t(4;11) Thomas M, Gessner A, Vornlocher HP, Hadwiger P, Greil J, chromosomal rearrangement exhibits B lineage and mono- Heidenreich O. (2005). Targeting MLL-AF4 with short cytic characteristics. Blood 65: 21–31. interfering RNAs inhibits clonogenicity and engraftment of Strasser A, Harris AW, Bath ML, Cory S. (1990). t(4;11)-positive human leukemic cells. Blood 106: 3559–3566. Novel primitive lymphoid tumours induced in transgenic Xia ZB, Popovic R, Chen J, Theisler C, Stuart T, Santillan DA mice by cooperation between and bcl-2. Nature 348: et al. (2005). The MLL fusion gene, MLL-AF4, regulates 331–333. cyclin-dependent kinase inhibitor CDKN1B (p27kip1) ex- Takeda S, Chen DY, Westergard TD, Fisher JK, Rubens JA, pression. Proc Natl Acad Sci USA 102: 14028–14033. Sasagawa S et al. (2006). Proteolysis of MLL family proteins Yu BD, Hess JL, Horning SE, Brown GA, Korsmeyer SJ. is essential for Taspase1-orchestrated cell cycle progression. (1995). Altered Hox expression and segmental identity in Genes Dev 29: 2397–2409. Mll-mutant mice. Nature 378: 505–508. Taranger CK, Noer A, Sorensen AL, Hakelien AM, Boquest Zhang J, Wang X, Li M, Han J, Chen B, Wang B et al. (2006). AC, Collas P. (2005). Induction of dedifferentiation, NANOGP8 is a retrogene expressed in . FEBS J 273: genomewide transcriptional programming, and epigenetic 1723–1730.

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

Oncogene