Alterations of the CxxC domain preclude oncogenic activation ... - Nature

Oncogene (2009) 28, 815–823

& 2009 Macmillan Publishers Limited All rights reserved 0950-9232/09 $32.00 www.nature.com/onc

ORIGINAL ARTICLE

Alterations of the CxxC domain preclude oncogenic activation of mixed-lineage leukemia 2 C Bach, D Mueller, S Buhl, MP Garcia-Cuellar and RK Slany Department of Genetics, University of Erlangen, Erlangen, Bavaria, Germany

The mixed-lineage leukemia (MLL) family of histone methyltransferases has become notorious for the participation of the founding member, MLL, in fusion proteins that cause acute leukemia. Despite structural conservation, no other MLL homolog has so far been found in a similar arrangement. Here, we show that fusion proteins based on Mll2, the closest relative of MLL, are incapable of transforming hematopoietic cells. Elaborate swap experiments identified the small CxxC zinc-binding region of Mll2 and an adjacent ‘post-CxxC’ stretch of basic amino acids as the essential determinants for the observed difference. Gel shift experiments indicated that the combined CxxC and post-CxxC domains of MLL and Mll2 possess almost indistinguishable DNA-binding properties in vitro. Within the cellular environment, however, these motifs guided MLL and Mll2 to a largely nonoverlapping target gene repertoire, as evidenced by nuclear localization, reporter assays, and measurements of homeobox gene levels in primary cells expressing MLL and Mll2 fusion proteins. Therefore, the CxxC domain appears to be a promising target for therapies aimed at MLL fusion proteins without affecting the general function of other MLL family members. Oncogene (2009) 28, 815–823; doi:10.1038/onc.2008.443; published online 8 December 2008 Keywords: CxxC domain; MLL; MLL2; transformation

Introduction Efficient transcription requires the cooperative action of many molecular machines that render the tightly packaged chromatin accessible to RNA synthesis. This process requires covalent histone modifications that are introduced by specialized enzymes. A subgroup of these proteins adds site-specific methyl groups to histones using the catalytic action of a so-called suppressor of variegation, enhancer of zeste, trithorax (SET) domain. Correspondence: Professor RK Slany, Department of Genetics, University Erlangen, Staudtstrasse 5, Building A2, Erlangen, 91058 Bavaria, Germany. E-mail: [email protected] Received 25 June 2008; revised 9 October 2008; accepted 1 November 2008; published online 8 December 2008

A special subclass of these SET proteins is responsible for the methylation of lysine 4 in histone H3. This modification is characteristically concentrated around the transcription start site of many active genes (Ruthenburg et al., 2007). The first member of this particular methyltransferase family has been originally discovered because of its involvement in leukemogenesis and hence was accordingly designated as mixed-lineage leukemia (MLL) protein (Krivtsov and Armstrong, 2007). Later, several other H3K4 methyltransferases (MLL2, MLL3 and MLL4; nomenclature according to Glaser et al., 2006) were found that shared homology with MLL. MLL2 (also sometimes designated as MLL4) is the gene most closely related to MLL. As indicated by the genomic microenvironment, MLL2 arose by a chromosomal duplication during mammalian evolution (FitzGerald and Diaz, 1999). MLL and MLL2 share the same overall structure and encode essentially identical functional domains (Glaser et al., 2006), and both proteins reside in macromolecular complexes of nearly identical composition (Dou et al., 2006). Despite the high similarity, only MLL has been described to date as a proto-oncogene that can be activated by chromosomal translocations. These events delete 30 portions of MLL, including the methyltransferase function and fuse the remaining 50 part with a variety of partner genes on other chromosomes. The corresponding chimeric MLL fusion proteins potently transform hematopoietic precursor cells. On a molecular level, MLL fusions inhibit hematopoietic differentiation by enforcing the continued expression of genes that are normally under control of unaltered MLL and that have to be mandatorily downregulated to allow maturation. With regard to hematopoietic development, the Hox homeobox genes such as Hoxa9 and Hoxa7 and their dimerization partner, Meis1, are the most important targets of MLL fusion proteins (Zeisig et al., 2004). Recent data indicate that several MLL fusion proteins can either stimulate transcriptional elongation (Mueller et al., 2007), or induce histone arginine methylation (Cheung et al., 2007) with the net effect of target activation. These fusion partner-associated activities are specifically recruited to the cognate targets by the action of the remaining MLL portion. Structure–function analysis identified two separate domains within the truncated MLL portion that are essential for the oncogenic capacity (Slany et al., 1998; Ayton et al., 2004). At the very N-terminus, MLL contains a menin

The CxxC domain of Mll2 precludes oncogenic activation C Bach et al

816

interaction domain. Menin was originally identified as the product of the tumor suppressor gene multiple endocrine neoplasia. Menin binding is absolutely essential for the function of the fusion protein, and menin is a component of a macromolecular complex that also contains normal MLL (Yokoyama et al., 2005; Caslini et al., 2007). Further downstream, a CxxC motif in MLL binds to unmethylated CpG DNA dinucleotides (Birke et al., 2002). Interestingly, the same region has also been described as a potential repression domain that recruits histone deacetylases, and it has been speculated that the CxxC domain might modulate the functional activator/repressor status of MLL and its derivatives (Zeleznik-Le et al., 1994; Xia et al., 2003). Despite the fact that both regions are highly conserved in MLL2, strikingly MLL2 has never been shown to be involved in leukemogenic fusion proteins. To study the functional relationship between MLL and MLL2, we wanted to elucidate whether MLL2 also could be converted to a leukemogenic fusion. Furthermore, we wanted to know which structural determinants of MLL and MLL2 might define the differences between the two proteins, if any. These are medically important questions because MLL-induced leukemias are particularly aggressive and undetected MLL2-fusion patients might benefit from an improved molecular diagnostics to allow a more intense therapy. Here, we show through exhaustive swap experiments that MLL2 has lost the intrinsic capability to form transforming fusions because of small differences in the MLL2-CxxC domain. These alterations most likely alter target gene specificity, and therefore this determinant within the MLL protein is a potential ‘soft spot’ for specific therapies. Results MLL and MLL2 are structurally similar but not functionally equivalent An overall sequence alignment between MLL (NM_005933) and MLL2 (NM_014727, also designated as MLL4) showed that the major functional domains of MLL are also conserved in MLL2 (Glaser et al., 2006). This is true in particular for the N-terminus where not only the domain organization but also the spacing between the respective motifs has been preserved (see Figure 1a for a schematic overview). The menin-binding domain, three AT-hook sequences and the CxxC motif that binds to nonmethylated CpG dinucleotides are all present at similar positions in MLL and in MLL2. The sequence homology is highest within these motifs, as exemplified by the alignment of the CxxC domain of MLL with human and mouse MLL2/Mll2 (Figure 1b). Almost all residues that are known to contact DNA (Allen et al., 2006) are identical in these proteins, and cross-species conservation reaches a maximum. To test whether this structural resemblance also extends to function, an artificial Mll2 fusion protein was constructed mimicking the highly transforming mixed lineage leukemia eleven nineteen leukemia (MLLENL) protein found in acute leukemias. As the Oncogene

cDNA of human MLL2 was not available, the mouse Mll2 cDNA (NM_029274, also known as tryptophantryptophan domain (WW) domain-binding protein 7) was used for these experiments. Overall, the human and mouse MLL2 proteins are approximately 90% identical and human as well as murine MLL fusion proteins are equally leukemogenic. Human MLL fusions have been tested extensively in vitro and in vivo in retroviral transduction/transplantation experiments, whereas murine Mll sequences have been joined with fusion partners in knock-in models of MLL (Chen et al., 2008). Most likely, therefore, differing properties of MLL and Mll2 are a consequence of protein variation and not of species divergence. The oncogenic capability of the respective MLL/Mll2 fusion proteins was assessed by an in vitro transformation assay that measures the capacity to block the differentiation of primary hematopoietic cells (Lavau et al., 1997). Unexpectedly, and despite appropriate expression, Mll2ENL did not yield a positive readout in these experiments indicating that notwithstanding the structural relatedness, there must be fundamental functional differences between MLL and Mll2 (Figure 2). The CxxC domain is responsible for the functional divergence of MLLENL and Mll2ENL To elucidate the motifs in Mll2 that are responsible for the observed functional difference, a series of swap experiments was carried out (Figure 3a). A set of Mll2/ MLL chimeric fusion proteins was constructed exchanging the C-terminal portion of Mll2 including the CxxC domain by corresponding sequences from MLL. This switch endowed the corresponding Mll2 fusion proteins with a significant transforming activity. Obviously, the menin-binding site and the AT-hook region of Mll2 were sufficiently conserved to replace the function of the corresponding regions of MLL. The proximal limit of the minimal MLL domain necessary to induce a transforming potential in Mll2ENL could be mapped to the N-terminus of the conserved CxxC core motif (Figure 3a, swaps 1–6). The slight rise in colonyforming capacity that was observed with progressively shorter swap inserts could be explained by an increase of the viral titer that was inversely proportional to the insert size (not shown). The downstream boundary of the minimal necessary MLL unit, however, extended considerably beyond the core CxxC domain, including a ‘post-CxxC’ region, characterized by a stretch of amino acids with a high frequency of basic residues, but little actual sequence conservation between MLL and Mll2 (Figure 3a, swaps 7–9, see also Figure 1b). To exclude the possibility that the transformation potential of Mll2ENL might simply be blocked by Mll2 inhibitory sequences outside the CxxC region, an Mll2ENL construct with an internal deletion upstream of the CxxC motif was tested (Figure 3a, swap # 10). As this construct did not score positive in the transformation assay, the CxxC domain itself must be responsible for the lack of transformation potential in Mll2ENL. The contributions of the core CxxC and the basic post-CxxC domains were assessed individually in further experi-


817

Figure 1 Structural homology of mixed-lineage leukemia (MLL) and MLL2. (a) Schematic alignment of MLL and MLL2. The figure is drawn approximately to scale. Numbers denote amino acid residues. Major functional domains are indicated as follows: menin, Nterminal menin interaction domain; AT-hooks, minor groove DNA-binding motif; CxxC, domain binding to CpG dinucleotides; PhD, plant homeodomain recognizing acetylated histones; SET, methyltransferase catalytic core. Here and throughout all figures, MLL is color coded in red, MLL2/Mll2 in green. (b) Alignment of the core-and post-CxxC domains of MLL (accession number NM_005933), human MLL2 (NM_014727) and murine Mll2 (NM_029274). Red boxes mark the cysteine-rich core involved in coordinating zinc. Basic amino acids are highlighted in blue. Green dots label amino acids that have been shown to be in direct contact with DNA (Allen et al., 2006). Residues identical in all proteins are underlined.

ments (Figure 3a, swaps 11 and 12). Interestingly, the insertion of a composite CxxC domain consisting of an Mll2 core and an MLL post-CxxC section or vice versa did slightly increase the transformation potential of the respective constructs emphasizing an individual contribution of the core- and post-CxxC domains toward the overall function. The CxxC domains of MLL and Mll2 have nearly identical in vitro DNA-binding properties The CxxC domain of MLL has been described as a CpG dinucleotide recognition motif that discriminates against methylated DNA sequences (Birke et al., 2002). Therefore, it was tested in electrophoretic mobility shift assay (EMSA) experiments whether the functional disparity between MLL and Mll2 might be due to differing DNA-binding properties. For this purpose, glutathione-S-transferase proteins encompassing the core- and post-CxxC moieties of MLL and

Mll2 were bacterially expressed and purified (Figure 4a). DNA-binding was assessed in mobility shifts with a probe that has been shown to be an efficient substrate for the MLL-CxxC domain (Birke et al., 2002). This 52-bp oligonucleotides contained six regularly spaced CpG dinucleotides. As controls, oligonucleotides with the same sequence but cytosine exchanged against 5-methyl cytosine (Figure 4b) and oligonucleotides of the same overall base composition but with CpG sequences reversed to guanine-cytosine dinucleotides were used (Figure 4c). Within normal experimental variations, the Mll2-CxxC DNA-binding properties were indistinguishable from MLL-CxxC in these experiments. Like MLL-CxxC, Mll2-CxxC was also dependent on nonmethylated CpG dinucleotides for efficient association with DNA. Therefore, the functional differences between these domains could not be explained by an inherently different DNAbinding mode. Oncogene


818

domain needed to be included. The different target spectrum of Mll2 fusions also translated into an inability to induce a hematopoietic differentiation block. Only MLLENL and swap 3 transduced cells showed signs of a maturation arrest after 5 days of in vitro culture with a high percentage of cells retaining the precursor specific marker c-kit, whereas Mll2ENLtransduced cells rapidly lost this surface molecule (Figure 5d).

Discussion Figure 2 Functional divergence of MLLENL and Mll2ENL. Western blot analysis of flag-tagged MLLENL and Mll2ENL expressed in retroviral packaging cells (left panel). Representative result of a replating assay (right panel). Primary hematopoietic cells were transduced with the expression construct indicated on top and serially replated in methylcellulose medium. Photographs show stained colonies after two rounds of replating. Colony formation is a consequence of a differentiation block that allows persistence of self-renewable precursor cells. A schematic cartoon of the construct setup is shown at the bottom.

The CxxC domain determines subnuclear localization and target gene selection of MLL/Mll2 fusion proteins A reason for the observed functional disparity of MLLENL and Mll2ENL might be that these proteins recognize different target genes. If this were the case, both proteins should be localized at least in part in different subnuclear compartments. Therefore, the nuclear localization of the two fusion proteins was determined in living cells by joining the respective Ntermini of MLL with green fluorescent protein (GFP) and of Mll2 with red fluorescent protein (RFP). Indeed, areas of limited colocalization for MLL-GFP and Mll2RFP were observed indicating the possibility that MLL and Mll2 do not share all target loci (Figure 5a, upper panel). To control for artifacts that might be introduced by the differing oligomerization behavior of GFP versus RFP, the same type of experiment was repeated with essentially the same result employing dTomato, a dimerizing derivative of tetrameric RFP (Figure 5a, lower panel). Next, the mRNA abundance of three known (Zeisig et al., 2004) MLLENL target genes (Hoxa7, Hoxa9 and Meis1) was measured by quantitative reverse transcriptase PCR in primary cells transduced with MLLENL, Mll2ENL or swap; 3 (Figure 5b). Only MLLENL and to a lesser extent swap 3 were able to activate the expression of these three genes that are known to be essential for transformation. In a similar direction, Mll2ENL could not activate a luciferase construct under control of the Hoxa7 promotor (Schreiner et al., 1999) in a transient reporter assay (Figure 5c, left panel). Similar to the results obtained in primary cells, an exchange of the Mll2-CxxC domain against the corresponding MLL moiety rescued a significant part of the transient transactivation activity (Figure 5c, right panel). Transient transactivation and transformation capability was generally correlated with the exception of swap 5 indicating that for optimal transactivation sequences upstream of the core CxxC Oncogene

In this report, we present evidence that Mll2, the closest homolog of the highly leukemogenic proto-oncoprotein, MLL, cannot be converted to a transforming protein because of an intrinsic functional difference encoded by the structurally conserved CxxC domain. This was unanticipated because MLL and MLL2 not only share major domains, but both proteins are also the members of a macromolecular complex with identical composition. Both MLL and MLL2 are associated with WDR5, RbBP5, HCF1/2, MENIN and ASH2 (Wysocka et al., 2003; Hughes et al., 2004; Yokoyama et al., 2004; Dou et al., 2005) ,and both proteins catalyze the trimethylation of histone H3 at lysine 4 (H3K4) to enable efficient transcriptional initiation. Therefore, it is a long-standing question why MLL is frequently involved in chromosomal translocations creating leukemogenic fusion proteins, whereas MLL2 has never been found in a comparable situation. Rare examples of MLL2 amplification in solid tumors are the only known instances where it is altered in neoplastic disease (Huntsman et al., 1999). The experiments shown here reveal that even if the MLL2 gene were the target of chromosomal translocations, the resulting fusion protein would not be transforming. Surprisingly, this is due to evolutionary changes in a small domain, which overall and in comparison with the remainder of the protein, shows high sequence conservation. This CxxC domain, a zinccoordinating motif, is present in several proteins where it serves as a module that binds to CpG dinucleotides (Lee et al., 2001). In MLL as well as in MLL2, it also discriminates against methylation presumably targeting the proteins to active promoters with unmethylated CpG islands. The structure of the CxxC fold has been solved; however, it remains largely enigmatic how a more specific target site selection might be achieved (Allen et al., 2006). One report describes the recruitment of the MLL2 complex to the b-globin locus by the ‘nuclear factor-erythroid derived 2’ transcription factor in erythroid cells (Demers et al., 2007), but it was not determined which component of the complex was responsible for the direct interaction with nuclear factor-erythroid derived 2. In addition to DNA, the region around the CxxC domain also may interact with various repressor proteins. Histone deacetylases 1 and 2, as well as the polycomb repressive complex members BMI-1 and HPC2 coprecipitate in vitro with recombinant portions of MLL that contain the CxxC motif (Xia


819

Figure 3 Core-and post-CxxC domains are responsible for the difference between mixed-lineage leukemia (MLL) and Mll2. (a) Results of bone marrow replating experiments for a series of swap clones. Numbers above the respective construct correspond to MLL amino acids. Numbers below indicate Mll2 residues. Functional domains are labeled as in Figure 1. The bar graph gives the average and standard deviation of third round colonies determined in triplicates. To allow a valid comparison across biological replicates with different bone marrow donors, colony numbers are given in relative units with the result for the canonical MLLENL construct set to 1.0 (equals 100%). The absolute colony number for MLLENL in the third round of plating ranged from approximately 50 to 1500 colonies per 10 000 cells seeded, depending on the respective batch of hematopoietic precursor cells used for transduction. (b) Expression of swap constructs 1–12, as detected by antiflag western blot. The lane numbers correspond to the respective swap constructs shown in panel a.

et al., 2003). It is not completely clear whether the evolutionary changes that lead to the divergence of Mll2/MLL2 altered target site selection, repression properties or both functions of the CxxC region. The differences in the nuclear distribution of Mll2ENL versus MLLENL argue for the first possibility. Moreover, gene expression studies showed that Mll2ENL can work as transcriptional activator, because expression of this fusion in primary bone marrow cells induced the myeloid developmental gene Mef2c (not shown). Altered binding properties for repressor proteins might additionally thwart the capability of Mll2ENL to activate a target gene set similar to MLLENL that would cause transformation of hematopoietic cells. Knockout studies also confirm a divergent role of MLL and MLL2. Both individual knockouts are embryonal lethal indicating that Mll and Mll2 are not able to substitute each other throughout embryonic

development (Yu et al., 1995; Glaser et al., 2006). Here, we pinpoint for the first time a functional unit inside these proteins that might be responsible for the observed difference and it might be interesting to see whether swapping solely the CxxC domain indeed converts the identity of the whole protein. In any case, this domain is a rewarding object for further studies, because a closer understanding of its mode of target selection might open the possibility to interfere with the activation of downstream targets by MLL fusion proteins.

Materials and methods Plasmids and antibodies All MLL and Mll2 fusion protein derivatives were constructed with an N-terminal flag tag and inserted into the retroviral vector pMSCV (Clontech, Palo Alto, CA, USA). The Oncogene


820

Figure 4 The in vitro DNA-binding properties of MLL-CxxC and Mll2-CxxC are similar. (a) Coomassie-stained gel of the purified glutathione-S-transferase CxxC fusion proteins as used in the electrophoretic mobility shift assay (EMSA) experiments. Numbers denote the residues of MLL and Mll2, respectively, that were included in the protein. (b) Binding of MLL-CxxC and Mll2-CxxC to CpG and methyl-CpG containing DNA. The indicated amounts of protein were incubated with a radioactively labeled 52-bp oligonucleotides containing six regularily spaced CpG dinucleotides that has been found to be a high affinity template for the MLLCxxC domain (Birke et al., 2002). Oligonucleotides of identical sequence but with cytosines in CpG changed to 5-methyl cytosin (CpGCH3) served as control. Glutathione-S-transferase did not bind to these DNA probes (Birke et al., 2002 and not shown). (c) Analogous experiment as in panel b employing either the CpG oligonucleotides for control or oligonucleotides with identical sequence composition but with all CpG sequences reversed to GpC dinucleotides.

MLLENL sequence was reported by Tkachuk et al. (1992). The nucleotide sequence of Mll2 is accessible in the database under NM_029274. All Mll2 portions were amplified from a cDNA kindly provided by F Stewart (Dresden, Germany) and the resulting clones were checked by sequencing. For Mll2/MLL swap constructs, a unique ZraI restriction site located at nt 3331 of the Mll2 cDNA sequence was chosen as joining point. For expression and purification of glutathione fusion proteins, the pGEX vector system (GE Healthcare, Munich, Germany) was used. MLL and Mll2 fusions with either GFP or RFP were performed in pEGFPN1 and pDSRed1-N1 from Clontech (Palo Alto, CA, USA). The dTomato sequence was derived from pLeGO-dTomato (Weber et al., 2008). Oncogene

Anti-flag (M2) antibody was from Sigma, (Taufkirchen, Germany) and antibodies for fluorescence-activated cell sorting analysis were purchased from BDBiosciences (San Jose, CA, USA). MLL fusion protein transformation assay The transformation capacity of MLL/Mll2 fusion protein derivatives was determined by serial replating assays as reported earlier(Lavau et al., 1997). In short, hematopoietic progenitor cells from mouse bone marrow were transduced with the respective MLL construct. Normal cells mature and stop proliferation after repeated replating in a semisolid medium, whereas transformed cells experience a block in


821

Figure 5 The CxxC domain is a determinant for target gene recognition. (a) Nuclear localization of MLL fused to green fluorescent protein (GFP) and Mll2 fused to red fluorescent protein (RFP) proteins in living cells (upper panel). HEK293 cells were transfected with the constructs indicated. Nuclear localization was recorded with a Zeiss Axioskop fitted with a Nikon DS-FII (5 Mpx) color camera. An electronic overlay shows a largely nonoverlapping distribution of the two proteins. The same experiment was repeated with MLL-GFP and Mll2-dTomato (lower panel). (b) Quantitative reverse transcriptase PCR (qRT–PCR) measurement of the known MLLENL target genes, Meis1, Hoxa7 and Hoxa9, in primary hematopoietic cells 5 days after transduction with either MLLENL, Mll2ENL or swap 3. Bar graph shows average values and standard deviations of triplicate experiments. (c) Transient reporter assay with a Hoxa7 promoter-driven luciferase construct coelectroporated with increasing amounts of MLLENL or Mll2ENL into REH pre-B cells (left panel). Given are relative luciferase outputs with the basal level in the absence of additional MLL/Mll2 fusion proteins arbitrarily set to one. The bar graph charts averages and standard deviations of triplicates. A selection of swap clones was tested for the capability to transactivate the Hoxa7 promoter in transient transfection assays (right panel). Constant amounts of expression constructs (0.9 mg) were coelectroporated with 0.1 mg Hoxa7 reporter into REH cells. For better comparability, all luciferase outputs were normalized to the fMll2ENL plus reporter experiment. Average and standard deviations of triplicates are given. Numbers correspond to the clone designations in Figure 3 and the transforming capacity of the construct is indicated by ‘ þ ’ and ‘’, respectively. (d) Surface c-kit (CD117) levels of primary hematopoietic cells determined by flow cytometry 5 days after transduction with the indicated constructs. Numbers denote the percentage of cells within the drawn counting region. Oncogene


822 differentiation and form colonies. Therefore, transformation can roughly be quantitated by colony numbers. Replating assays were carried out at least three times per construct. Colonies were stained with p-iodonitrotetrazolium violet (INT, Sigma, Taufkirchen, Germany), photographed and counted. To allow comparison across biological replicates with different bone marrow donors, all colony counts were expressed relative to the colony number achieved with the positive control MLLENL. EMSA, quantitative reverse transcriptase PCR and luciferase assays Electrophoretic mobility shift assay experiments were performed with glutathione-S-transferase fusion proteins containing the CxxC domains of MLL and Mll2 as indicated. The EMSA procedure has been described in detail (Birke et al., 2002). Quantative reverse transcriptase PCR was performed to determine mRNA levels of Meis1, Hoxa7 and Hoxa9 in a SYBR green-based assay using following primers: Meis1 (primer fw 50 -CCTCTGCACTCGCATCAGTAC-30 ; primer rev 50 -GTTTGGCGAACACCGCTATATC-30 ); Hoxa7 (primer fw 50 -CCCTTCGCGTCCGGCTATG-30 , primer rev 50 -GTCTGG CGTCCCCGCTTCC-30 ); Hoxa9 (primer fw 50 -GCTCTCCTT CGCGGGCTTACC-30 , primer rev 50 -GGGCATCGCTTCTTC CGAGTG-30 ). All samples were normalized to b-actin (primer fw

50 -CCAACTGGGACGACATGGAG-30 , primer rev 50 -CTCGT AGATGGGCACAGTGTG-30 ). RNA for these experiments was isolated from primary hematopoietic precursor cells isolated from bone marrow. The cells were transduced with the respective viral construct and expanded for 5 days in methocel medium with cytokines (IL-3, IL-6, GM-CSF 10 ng/ml; SCF 100 ng/ml) under appropriate selection. RNA was isolated with RNeasy kits from Quiagen (Hilden, Germany) according to the instructions of the manufacturer. Luciferase assays were carried out as described (Schreiner et al., 1999) with REH pre-B cells using a Hoxa7 promoter driven luciferase reporter and pMSCV-MLLENL or pMSCVMll2ENL constructs in varying concentrations as expression plasmids.

Acknowledgements We thank Bernd Zeisig and Silvio Scheel for help in early stages of this work. Technical support of Renate Zimmermann and sharing of reagents by Francis Stewart, Boris Fehse and Carol Stocking is gratefully acknowledged. RKS was supported by DFG Grant SL27/6–2 and in part by SFB473/D2. Equipment funding came from Jose-Carerras-Stiftung and Curt-Bohnewald-Fond.

References Allen MD, Grummitt CG, Hilcenko C, Min SY, Tonkin LM, Johnson CM et al. (2006). Solution structure of the nonmethyl-CpG-binding CXXC domain of the leukaemia-associated MLL histone methyltransferase. EMBO J 25: 4503–4512. Ayton PM, Chen EH, Cleary ML. (2004). Binding to nonmethylated CpG DNA is essential for target recognition, transactivation, and myeloid transformation by an MLL oncoprotein. Mol Cell Biol 24: 10470–10478. Birke M, Schreiner S, Garcia-Cuellar MP, Mahr K, Titgemeyer F, Slany RK. (2002). The MT domain of the proto-oncoprotein MLL binds to CpG-containing DNA and discriminates against methylation. Nucleic Acids Res 30: 958–965. Caslini C, Yang Z, El-Osta M, Milne TA, Slany RK, Hess JL. (2007). Interaction of MLL amino terminal sequences with menin is required for transformation. Cancer Res 67: 7275–7283. Chen W, Kumar AR, Hudson WA, Li Q, Wu B, Staggs RA et al. (2008). Malignant transformation initiated by Mll-AF9: gene dosage and critical target cells. Cancer Cell 13: 432–440. Cheung N, Chan LC, Thompson A, Cleary ML, So CW. (2007). Protein arginine-methyltransferase-dependent oncogenesis. Nat Cell Biol 9: 1208–1215. Demers C, Chaturvedi CP, Ranish JA, Juban G, Lai P, Morle F et al. (2007). Activator-mediated recruitment of the MLL2 methyltransferase complex to the beta-globin locus. Mol Cell 27: 573–584. Dou Y, Milne TA, Ruthenburg AJ, Lee S, Lee JW, Verdine GL et al. (2006). Regulation of MLL1 H3K4 methyltransferase activity by its core components. Nat Struct Mol Biol 13: 713–719. Dou Y, Milne TA, Tackett AJ, Smith ER, Fukuda A, Wysocka J et al. (2005). Physical association and coordinate function of the H3 K4 methyltransferase MLL1 and the H4 K16 acetyltransferase MOF. Cell 121: 873–885. FitzGerald KT, Diaz MO. (1999). MLL2: A new mammalian member of the trx/MLL family of genes. Genomics 59: 187–192. Glaser S, Schaft J, Lubitz S, Vintersten K, van der Hoeven F, Tufteland KR et al. (2006). Multiple epigenetic maintenance factors implicated by the loss of Mll2 in mouse development. Development 133: 1423–1432. Hughes CM, Rozenblatt-Rosen O, Milne TA, Copeland TD, Levine SS, Lee JC et al. (2004). Menin associates with a trithorax family Oncogene

histone methyltransferase complex and with the hoxc8 locus. Mol Cell 13: 587–597. Huntsman DG, Chin SF, Muleris M, Batley SJ, Collins VP, Wiedemann LM et al. (1999). MLL2, the second human homolog of the Drosophila trithorax gene, maps to 19q13.1 and is amplified in solid tumor cell lines. Oncogene 18: 7975–7984. Krivtsov AV, Armstrong SA. (2007). MLL translocations, histone modifications and leukaemia stem-cell development. Nat Rev Cancer 7: 823–833. Lavau C, Szilvassy SJ, Slany R, Cleary ML. (1997). Immortalization and leukemic transformation of a myelomonocytic precursor by retrovirally transduced HRX-ENL. EMBO J 16: 4226–4237. Lee JH, Voo KS, Skalnik DG. (2001). Identification and characterization of the DNA binding domain of CpG-binding protein. J Biol Chem 276: 44669–44676. Mueller D, Bach C, Zeisig D, Garcia-Cuellar MP, Monroe S, Sreekumar A et al. (2007). A role for the MLL fusion partner ENL in transcriptional elongation and chromatin modification. Blood 110: 4445–4454. Ruthenburg AJ, Allis CD, Wysocka J. (2007). Methylation of lysine 4 on histone H3: intricacy of writing and reading a single epigenetic mark. Mol Cell 25: 15–30. Schreiner SA, Garcia-Cuellar MP, Fey GH, Slany RK. (1999). The leukemogenic fusion of MLL with ENL creates a novel transcriptional transactivator. Leukemia 13: 1525–1533. Slany RK, Lavau C, Cleary ML. (1998). The oncogenic capacity of HRX-ENL requires the transcriptional transactivation activity of ENL and the DNA binding motifs of HRX. Mol Cell Biol 18: 122–129. Tkachuk DC, Kohler S, Cleary ML. (1992). Involvement of a homolog of Drosophila trithorax by 11q23 chromosomal translocations in acute leukemias. Cell 71: 691–700. Weber K, Bartsch U, Stocking C, Fehse B. (2008). A multicolor panel of novel lentiviral ‘gene ontology’ (LeGO) vectors for functional gene analysis. Mol Ther 16: 698–706. Wysocka J, Myers MP, Laherty CD, Eisenman RN, Herr W. (2003). Human Sin3 deacetylase and trithorax-related Set1/Ash2 histone H3-K4 methyltransferase are tethered together selectively by the cell-proliferation factor HCF-1. Genes Dev 17: 896–911.


823 Xia ZB, Anderson M, Diaz MO, Zeleznik-Le NJ. (2003). MLL repression domain interacts with histone deacetylases, the polycomb group proteins HPC2 and BMI-1, and the corepressor C-terminalbinding protein. Proc Natl Acad Sci USA 100: 8342–8347. Yokoyama A, Somervaille TC, Smith KS, Rozenblatt-Rosen O, Meyerson M, Cleary ML. (2005). The menin tumor suppressor protein is an essential oncogenic cofactor for MLL-associated leukemogenesis. Cell 123: 207–218. Yokoyama A, Wang Z, Wysocka J, Sanyal M, Aufiero DJ, Kitabayashi I et al. (2004). Leukemia proto-oncoprotein MLL forms a SET1-like histone methyltransferase complex with menin to regulate Hox gene expression. Mol Cell Biol 24: 5639–5649.

Yu BD, Hess JL, Horning SE, Brown GA, Korsmeyer SJ. (1995). Altered Hox expression and segmental identity in Mll-mutant mice. Nature 378: 505–508. Zeisig BB, Milne T, Garcia-Cuellar MP, Schreiner S, Martin ME, Fuchs U et al. (2004). Hoxa9 and Meis1 are key targets for MLL-ENL-mediated cellular immortalization. Mol Cell Biol 24: 617–628. Zeleznik-Le NJ, Harden AM, Rowley JD. (1994). 11q23 translocations split the ‘AT-hook’ cruciform DNA-binding region and the transcriptional repression domain from the activation domain of the mixed-lineage leukemia (MLL) gene. Proc Natl Acad Sci USA 91: 10610–10614.

Oncogene