c-FMS chromatin structure and expression in normal and ... - Nature

Oncogene (2005) 24, 3643–3651

& 2005 Nature Publishing Group All rights reserved 0950-9232/05 $30.00 www.nature.com/onc

c-FMS chromatin structure and expression in normal and leukaemic myelopoiesis George Alexander Follows*,1,2, Hiromi Tagoh1, Stephen John Richards2, Svitlana Melnik1, Helen Dickinson3, Erica de Wynter1, Pascal Lefevre1, Gareth John Morgan2 and Constanze Bonifer1 1

Molecular Medicine Unit, St James’s University Hospital, Leeds, UK; 2Academic Unit of Haematology and Oncology, Leeds General Infirmary, Leeds, UK; 3Department of Cytogenetics, St James’s University Hospital, Leeds, UK

The macrophage colony-stimulating factor receptor is encoded by the c-FMS gene, and it has been suggested that altered regulation of c-FMS expression may contribute to leukaemic transformation. c-FMS is expressed in pluripotent haemopoietic precursor cells and is subsequently upregulated during monocytic differentiation, but downregulated during granulopoiesis. We have examined transcription factor occupancy and aspects of chromatin structure of the critical c-FMS regulatory element located within the second intron (FIRE – fms intonic regulatory element) during normal and leukaemic myelopoiesis. Granulocytic differentiation from normal and leukaemic precursors is accompanied by loss of transcription factors at FIRE and downregulated c-FMS expression. The presence of AML1-ETO in leukaemic cells does not prevent this disassembly. In nonleukaemic cells, granulocytic differentiation is accompanied by reversal to a chromatin fine structure characteristic of cFMS-nonexpressing cells. In addition, we show that lowlevel expression of the gene in leukaemic blast cells and granulocytes does not associate with increased CpG methylation across the c-FMS locus. Oncogene (2005) 24, 3643–3651. doi:10.1038/sj.onc.1208655 Published onlline 4 April 2005 Keywords: c-FMS; AML; myelopoiesis; AML1-ETO

Introduction Understanding the mechanisms, which appropriately silence or activate genes, is critical to our understanding of both normal haematopoiesis and leukaemic transformation. It is increasingly recognized that epigenetic mechanisms mediated by alterations in transcription factor networks can have profound effects on gene *Correspondence: GA Follows. Current address: Department of Haematology, Cambridge Institute for Medical Research, Hills Road, Cambridge, UK; E-mail: [email protected] Received 15 November 2004; revised 22 February 2005; accepted 25 February 2005; published onlline 4 April 2005

expression. Deregulation of these networks can both prevent the normal differentiation of haematopoietic progenitor cells as well as contribute directly to the transformation of the cell. c-FMS encodes the 150 kDa tyrosine kinase receptor for macrophage colony-stimulating factor (M-CSF) that has been shown to play a critical role in the differentiation and maintenance of macrophages (Dai et al., 2002). The gene is the cellular counterpart of the feline sarcoma virus v-fms (Sherr et al., 1985). It was first linked to monocytic differentiation when transcripts were identified in peripheral blood monocytes, and monocytic, but not granulocytic, differentiation of HL60 cells was accompanied by upregulation of c-FMS expression (Sariban et al., 1985). Low levels of mRNA transcripts can be detected in haematopoietic progenitor cells, although cell surface expression of the protein does not become detectable until later in monocytic differentiation (Tagoh et al., 2002). Granulocytic differentiation of myeloid progenitor cells results in a marked downregulation of c-FMS expression (Lichtenberger et al., 1999). The oncogenic potential of c-FMS was originally demonstrated in a murine model of myelomonoblastic leukaemia (Gisselbrecht et al., 1987), although at present there is little convincing evidence linking overexpression of the gene and human leukaemia. In contrast with the FLT3 tyrosine kinase receptor (Kelly and Gilliland, 2002), mutations of the c-FMS gene are rare in AML patients (Springall et al., 1993), and gene expression analyses of bone marrow samples from AML patients have not found c-FMS to be overexpressed in a range of cytogenetic subgroups of AML (Debernardi et al., 2003; Follows et al., 2003a). Furthermore, a recent study has found an association between the AML-M2 subtype and reduced expression of c-FMS (Casas et al., 2003). Indeed, there is evidence that with certain leukaemic disorders, inactivation of the gene may contribute to leukaemic transformation (Boultwood et al., 1991). The human cFMS gene is located on chromosome 5 at band 5q33.3, and the gene is consistently lost in myelodysplasia patients with a partial deletion of chromosome 5. In addition, in as many as 40% of these patients, the

c-FMS expression in leukaemic transformation GA Follows et al

3644

gene is deleted from the apparently normal chromosome 5 (Boultwood et al., 1991). c-FMS deletions have now also been identified at lower frequencies in AML samples (McGlynn et al., 1997). There is evidence that the reduced c-FMS expression seen in AML may reflect inhibition of the normal regulatory elements, which are required for appropriate c-FMS expression. The coordinated assembly of transcription factors, such as PU.1 and AML1, at the c-FMS promoter and FIRE (fms intronic regulatory element) is required for appropriate expression of the gene (Tagoh et al., 2002; Follows et al., 2003b), and the deletion of FIRE in constructs introduced into macrophage cell lines and transgenic mice markedly reduces c-FMS expression (Himes et al., 2001; Sasmono et al., 2003). Recent experiments indicate that loss or disruption of key transcription factor function at the c-FMS locus parallels leukaemic transformation. Graded disruption of PU.1 expression in mice, with associated disruption of M-CSF-responsive pathways, has been shown to be leukaemogenic (Rosenbauer et al., 2004). Furthermore, we have previously shown that disruption of the AML1 transcription factor, by recruitment of AML1-ETO to the core regulatory element in FIRE, represses c-FMS mRNA expression (Follows et al., 2003a). Overexpression of AML1-ETO in U937 cells also leads to a reduction in c-FMS expression (Alcalay et al., 2003; Alcalay, 2003, personal communication) and inhibition of AML1-ETO with RNAi induces increased surface expression of the MCSF receptor in Kasumi-1 cells (Heidenreich et al., 2003). Here, we have studied transcription factor recruitment and chromatin structure of the FIRE element in normal monocytic and granulocytic differentiation and compared this with AML-M2 leukaemia samples with and without t(8;21). This translocation results in the expression of the aberrant transcription factor, AML1ETO. These subtypes of leukaemia characteristically have an FAB (French–American–British) M2 morphological phenotype, with minimal monocytic but partial dysplastic granulocytic differentiation. In the study presented here, we asked whether normal granulocytic differentiation is accompanied by a loss of transcription factor complexes at the c-FMS locus. We show that these complexes disassemble from FIRE with normal granulocytic differentiation and that this also occurs in leukaemic cells with and without AML-ETO. Normal CD34 þ bone marrow cells and CD34 þ leukaemic blasts have similar patterns of transcription factor recruitment to macrophages, which is independent of the presence of t(8;21). However, analysis of FIRE chromatin fine structure reveals differences between the leukaemic blasts and nonleukaemic cells. The former express low levels of c-FMS, but have a chromatin structure similar to monocyte/macrophages, which express high levels of the gene. Despite the low levels of c-FMS expression seen in both leukaemic blasts and mature granulocytes, CpG sites at crucial cis-regulatory sites at the c-FMS locus are not methylated. Oncogene

Results An outline map of the first 10 kb of the human c-FMS gene is presented in Figure 1a, showing the relative positions of the three regulatory elements: promoter, FIRE and FIRE-1 kb. These elements are highly conserved and have been shown to be DNaseI hypersensitive in mouse and human macrophages, which express high levels of c-FMS (Tagoh et al., 2002; Follows et al., 2003a). AML-M2 samples are characterized by a block in monocytic differentiation With the original FAB classification of AML, by definition, AML-M2 leukaemias must demonstrate partial granulocytic differentiation of bone marrow cells, with minimal monocytic differentiation. Over 90% of t(8;21) cases have AML-M2 morphology (Downing, 1999). Figure 2a shows a typical differentiation profile of bone marrow cells from a newly diagnosed AML-M2 patient with t(8;21). The undifferentiated leukaemic blasts have a characteristic phenotype: CD34 þ , CD13 þ , CD117 þ , HLA-DR þ , CD15 and CD64. The predominant cell type in the CD34 compartment is granulocytic, with a cell surface phenotype: CD15 þ , CD64 þ and CD14. There is a lack of monocytic differentiation, as shown by the low numbers of CD64 þ and CD14 þ cells. We have previously demonstrated that leukaemic AML-M2 blasts with or without t(8;21) express low levels of c-FMS (Follows et al., 2003a). This has subsequently been confirmed by other groups (Casas et al., 2003). Here, we show that in four patients with t(8;21), the level of c-FMS expression is lower in the CD34 þ cells compared with the differentiated CD34 fraction (Figure 2b). In all cases, the CD34 þ leukaemic fraction expressed lower levels of c-FMS compared with the mean level of c-FMS expression observed from normal bone marrow donor CD34 þ cells. Transcription factor complexes disassemble from FIRE with normal and leukaemic granulocytic differentiation We next asked whether the silencing of c-FMS in granulocytic differentiation is accompanied by a disassembly of transcription factor complexes on the cFMS locus. To obtain information about the transcription factor occupancy and chromatin fine structure, we used two different in vivo footprinting methods as described previously (Kontaraki et al., 2000). Bound transcription factors and chromatin structure affect the reactivity of DNA with dimethyl sulphate (DMS) or UV, both of which can be applied to intact cells. After in vivo formation of alkylated or dimerized bases, respectively, the position of these lesions is determined at nucleotide resolution by use of ligation-mediated polymerase chain reaction (LM-PCR) or a related technique (TD-PCR (Chen et al., 2001)). DMS footprinting indicates the position of transcription factor binding sites occupied by high-affinity DNA-binding proteins,


3645

Figure 1 (a) An outline map of the 10 kb of the human c-FMS locus spanning the promoter. Conserved regulatory elements are indicated with shaded boxes. P: promoter, F-1: FIRE-1 kb; F: FIRE. Numbered exons are indicated with open boxes. DNase Ihypersensitive sites in macrophages, which express high levels of c-FMS mRNA, are indicated with down arrows. (b) A map of the three regulatory elements identified previously (Himes et al., 2001; Tagoh et al., 2002; Follows et al., 2003b). Numbered CpG sites are identified with diamonds. The methylation-sensitive restriction enzyme sites used to quantify CpG methylation are indicated. Transcription factor binding sites are indicated with shaded boxes

but does not reveal information about general chromatin structure. UV footprinting highlights UV dimer formation between neighbouring bases (mostly pyrimidines) and generates information regarding DNA topology, which in addition can be used to visualize the action of general chromatin components indirectly. We have previously shown that the conserved occupied core of human FIRE consists of a central ets/SP1/AML1 element that is flanked by GC-rich sequences that may be binding sites for the transcription factor AP2 or MZF-1 (Morris et al., 1994). In addition, a conserved ets binding site that is located downstream of the core elements is footprinted in c-FMS-expressing cells. Figure 3a and b show a representative in vivo DMS-footprinting experiment of the lower and upper strands at FIRE, respectively. Transcription factors are clearly bound to the central core and flanking regions in both normal bone marrow CD34 þ cells and macrophages. In contrast, there is no evidence of

transcription factor binding to FIRE in either nonexpressing fibroblasts or granulocytes, where both lanes are highly similar to the in vitro DMS reaction (G). We have shown previously by chromatin immunoprecipitation (ChIP) assays that AML1-ETO binds the FIRE core element in t(8;21)-positive leukaemic blast cells and is part of an extended complex that forms a DNase I-hypersensitive site (Follows et al., 2003a). From previous ChIP experiments with t(8;21)-negative leukaemia cell lines, we have shown that the FIRE element binds AML1, with a DMS-footprinting pattern similar to that seen with t(8;21)-positive cell lines (Follows et al., 2003a). Here, we asked whether this complex is able to dissociate from its target sequence in t(8;21) granulocytes. Figure 4 shows an in vivo DMSfootprinting experiment looking at of the lower strand of FIRE with cells from the CD34 þ and CD34 fractions of leukaemic bone marrow. In all cases, the Oncogene


3646

Figure 2 (a) Multicolour flow cytometry demonstrates typical granulocytic differentiation seen in FAB M2 leukaemia bone marrow samples. This patient (P3) had t(8;21). The majority of the CD34 population are in R3, which are myeloid and predominantly granulocytic (CD64high/CD14low). The low number of cells in region R5 confirms the lack of monocytic differentiation in this t(8;21) leukaemia. (b) Normal granulocytes as well as CD34 þ (P1 þ , P2 þ , P3 þ , P4 þ ) and CD34 (P1, P2, P3, P4) cells from leukaemia patients with AML-M2 t(8;21) express low levels of c-FMS mRNA compared to peripheral blood monocytes/macrophages and normal bone marrow CD34 þ cells (NBM CD34 þ ). Relative mRNA expression levels in the indicated cell types were measured by real-time PCR as described in Materials and methods. The numbers indicate average relative mRNA levels from at least two independent RNA preparations and/or three independent real-time PCR experiments

core element is occupied in the leukaemic blasts, but not in the granulocytic (CD34) compartment. This is most clearly seen with the ratio of protection and enhancement of the G residues at position þ 4433 and þ 4432, respectively. The loss of this transcription factor occupancy in CD34 leukaemic cells confirms that AML1-ETO disassembles from this element with granulocytic differentiation. Oncogene

Chromatin fine structure of FIRE in leukaemic and normal CD34 þ cells has clear similarities with that of macrophages We have previously identified regions of the FIRE element where in vivo UV footprinting has revealed consistent changes in pyrimidine dimer formation across the FIRE element (Follows et al., 2003b).


3647

Figure 3 In vivo DMS footprinting of FIRE reveals extensive protections and enhancements of DMS reactivity on both DNA strands in normal bone marrow CD34 þ cells and macrophages, but not peripheral blood granulocytes and fibroblasts. (a) Extensive protections, particularly over the AML1/SP-1/Ets binding site, and the enhancement of G þ 4432 were observed on the lower strand. There are further protections over a GC-rich region at þ 4465 and enhancements around an ETS binding consensus site at þ 4516. (b) On the upper strand, there are protections of GC-rich regions, which flank the AML1/Sp1/Ets site, and in addition, there is a prominent enhancement of G þ 4434 in normal bone marrow CD34 þ cells and primary macrophages. These protections and enhancements are absent from granulocytes and fibroblasts. (c) The DNA sequence at FIRE with protections and enhancements indicated as white and black circles, respectively. Regions ‘Y’ and ‘Z’ are areas of altered UV dimer formation shown in Figure 5. Sequences conserved between mouse and man are indicated by a line, sequence deviations are displayed as (*). G: G reaction with naked DNA. The numbers indicate the nucleotide positions relative to the ATG codon

Figure 5a shows a representative UV-footprinting experiment of this region in leukaemic and nonleukaemic cells, with quantification presented in Figure 5b. Region ‘Z’ is highly conserved between mouse and human, while region ‘Y’ is not. The pattern of dimer formation over region ‘Z’ in c-FMS-nonexpressing/low-expressing fibroblasts, granulocytes and normal CD34 þ cells is strikingly similar to that seen

previously in both human and mouse nonexpressing cells (Follows et al., 2003b). In contrast, in this region, the leukaemic CD34 þ cells have a c-FMS chromatin structure similar to macrophages, which express high levels of c-FMS. Over region ‘Y’, the leukaemic and normal CD34 þ cells appear more similar, in contrast with the appearance of this region in macrophages. Oncogene


3648

Figure 4 In vivo DMS footprinting of FIRE (lower strand) reveals evidence of bound transcription factor complexes to the core FIRE element in CD34 þ leukaemic blasts, which are not present in the leukaemic CD34 granulocytic fraction. (a) With CD34 þ leukaemic blasts, extensive protections over the AML1/SP-1/Ets binding site are observed, with consistent protection of G þ 4433 and enhancement of G þ 4432. (b) With leukaemic cells from the CD34 granulocytic fraction, there is no evidence of alterations in DMS reactivity over the core FIRE element compared with the control G reaction. This is most clearly seen with the lack of enhancement of G þ 4432, yet relative enhancement of G þ 4433. The findings in the CD34 þ and CD34 fractions are independent of the presence of t(8;21). Protections and enhancements indicated as white and black filled circles, respectively. P1–P9 refer to leukaemia samples as indicated in Materials and methods

Methylation of c-FMS regulatory element DNA does not contribute to silencing of c-FMS expression in normal and leukaemic myelopoiesis DNA methylation is a postreplication modification that primarily affects cytosine residues in CpG doublets. Using methylation-sensitive restriction enzyme digestion followed by real-time PCR to quantify digestion, the methylation status of specific CpG sites at the c-FMS promoter, FIRE-1 kb and FIRE in various cell types was assayed. Figure 1b indicates the position of the CpG sites within each element. Figure 6 shows that the promoter is relatively methylated in normal CD34 þ bone marrow cells compared with the majority of leukaemic CD34 þ blast cells, and myelomonocytic differentiation of normal cells associates with CpG demethylation. In contrast, FIRE and FIRE-1 kb CpG sites are demethylated in all haematopoietic cells studied, both normal and leukaemic. Granulocytic differentiation did not associate with DNA methylation. In addition, the presence of AML1-ETO, which we have previously shown binds FIRE in both patient and Kasumi-1 cell line samples (Follows et al., 2003a), does not induce CpG methylation. We confirmed this by performing bisulphite sequencing of the FIRE element from Kasumi-1 and leukaemia samples, which showed that all 13 CpG sites across the 500 bp FIRE element remain unmethylated in t(8;21) samples (data not shown).

Discussion We have previously studied the c-FMS gene and identified regulatory elements in both the mouse and Oncogene

human genes critical to appropriate gene expression (Tagoh et al., 2002; Follows et al., 2003b). In this study, we show that the downregulation of c-FMS expression observed in granulopoiesis clearly associates with a loss of transcription factor assembly at FIRE. This correlates with alterations in chromatin fine structure in these cells but does not depend on, or result in, CpG methylation. The mechanism by which the transcription factor complex disassembles from FIRE during granulopoiesis remains unclear. Extinction of expression of specific transcription factors, such as PU.1, could be the cause (Dahl et al., 2003), or alternatively, post-translational modification of transcription factors could prevent their assembly into complexes on DNA. We have also previously shown c-FMS to be a target for the AML1-ETO oncoprotein (Follows et al., 2003a), with maximal recruitment to the intronic elements. Binding of AML1-ETO to FIRE inhibited the transactivation of the c-FMS promoter, but when bound, associated with profound DNaseI hypersensitivity (Follows et al., 2003a). This finding of a reorganized chromatin structure in AML1-ETO-expressing leukaemic cells has now also been confirmed in patient cells by the UV-footprinting experiments presented here. We have hypothesized that AML1-ETO could contribute to the block in monocytic differentiation observed in t(8;21) AML, by preventing the upregulation of appropriate c-FMS expression. Our results presented here show that this repressive AML1-ETO complex also has to be part of a larger transcription complex that disassembles in differentiating granulocytic cells. This is noteworthy as it implies that AML1-ETO complex assembly on a target locus is not permanent and that active inhibition of the gene in the presence of the


3649

Figure 5 (a) In vivo UV footprinting of the FIRE (lower DNA strand) from normal cells and leukaemic CD34 þ cells with and without t(8;21). Transcription factor binding sites are indicated on the left of the figure. Region ‘C’ indicates control bands used for the quantification presented in Figure 5b. Areas ‘a’ and ‘b’ are indicated to clarify points of quantification in Figure 5b. Regions ‘X’, ‘Y’ and ‘Z’ have been identified previously (Follows et al., 2003b). Region ‘Z’ is a region of reduced UV dimer formation in both human and mouse macrophages (Follows et al., 2003b). Leukaemic CD34 þ cells demonstrate a reduced dimer formation over this region similar to macrophages (open circles) in region Z. Region ‘Y’ is a region of reduced dimer formation in human, but not mouse macrophages (Follows et al., 2003b). Only the highly c-FMS expressing macrophages have reduced dimer formation in this region. (b) Quantification of the experiment presented in Figure 5a, with bands quantified relative to the total signal intensity of each lane (percentage). Patients P1 and P3 are t(8;21) þ , while P8 and P10 are t(8;21)

proper set of transcription factors may require the persisting presence of the bound oncoprotein. The evidence presented in this paper, and that from our earlier work, strongly suggests that at this locus AML1ETO inhibits the normal regulation of the gene in an open chromatin context. This notion is also supported by our DNA methylation studies. Differential methylation of CpG sites often associates with cell differentiation, and abnormal methylation and the epigenetic inheritance of new patterns of gene expression are increasingly recognized as oncogenic mechanisms underlying cellular transformation (Rountree et al., 2001). Our results show that that there is no CpG methylation of the c-FMS locus in patients with t(8;21) AML, despite the gene being an AML1-ETO target. This indicates that inhibition/downregulation of expression does not always require the gene to be assembled into a closed heterochromatin-like structure. This contrasts with the finding of other leukaemic oncoproteins at other loci where binding of the PML-RARa oncopro-

tein leads to epigenetic silencing via the recruitment of a DNA methyltransferase to specific gene targets (Di Croce et al., 2002). Clearly, the mechanisms that direct correct transcription factor interaction and function are highly complex. A recent study has shown that the absolute concentration of critical transcription factors may be more important with regard to malignant transformation than their mere presence or absence (Rosenbauer et al., 2004). Through graded reduction, rather than complete removal, of PU.1 expression in mouse bone marrow cells, Rosenbauer et al were able to induce leukaemic transformation. In these cells, reduced PU.1 expression appeared to have altered the balance of transcription factor assembly at the c-FMS promoter and/or FIRE preventing c-FMS expression. This correlated with a striking loss of c-FMS expression and M-CSF responsiveness in bone marrow cultures, resulting in a block in differentiation at the myeloid precursor state that over time developed into a myeloid Oncogene


3650

Figure 6 Relative cytosine methylation at three separate CpG sites at the promoter, FIRE-1 kb and FIRE measured by the reactivity of the methylation-sensitive restriction enzymes as described in Materials and methods. CpG sites at all three elements were demethylated in leukaemic CD34 þ blast cells from patients with and without t(8;21). In all but one sample (P6 þ ), the promoter CpG appeared significantly more demethylated in leukaemic CD34 þ cells compared with nonleukaemic CD34 þ cells from normal donors (CD34 þ NBM). Granulocytes (Gran) and macrophages (Mac) from normal donors were demethylated at each CpG site. KAS: Kasumi-1 cell line (t(8;21) þ ); P1 þ –P9 þ : CD34 þ cells from individual patients (FAB M2 either with or without t(8;21), as detailed in Materials and methods); Fb: fibroblasts

leukaemia. Interestingly, this block in differentiation could be reversed after the reintroduction of PU.1. This is similar to the reversal of the differentiation block observed in t(8;21) cells following the inhibition of AML1-ETO by RNAi (Heidenreich et al., 2003). Together with our findings, these results indicate possible avenues for therapeutic intervention that are aimed at the aberrant oncoprotein itself and may not require the reprogramming of all epigenetic marks in leukaemic cells.

P10) following ammonium chloride red cell lysis. The purity of each cell sorted fraction was >97% and column-purified fraction was >95%. Granulocytes were isolated from the peripheral blood of healthy donors using standard Ficoll techniques. Normal monocyte/macrophages were obtained from peripheral blood by culturing mononuclear cells in IMDM containing 10% foetal calf serum, and 50ng/ml MCSF (R þ D Systems). Adherent monocyte/macrophages were harvested after 24 h in culture. Human fibroblasts were cultured and harvested as described previously (Follows et al., 2003a). RT–PCR

Materials and methods Isolating patient and donor material Bone marrow samples were collected from consented patients when diagnostic bone marrow aspirates were performed. Morphological, flow cytometric, PCR and cytogenetic analyses were performed on all samples as components of the standard diagnostic work-up. Patients P1–P5 are positive for t(8;21), while patients P6–P10 are negative for t(8;21) by cytogenetic and PCR analysis. Normal CD34 þ cells were isolated from consented bone marrow donors. The CD34 þ and CD34 fractions were isolated using either cell sorting (Becton-Dickinson FACS Vantage) (normal donors and patients P1–P4 and P6–P8) or multiple column passes with a CD34 progenitor cell isolation kit (Miltenyi Biotec) (P5, P9, Oncogene

RNA was extracted and reverse transcribed using oligo(dT) primer and M-MLV reverse transcriptase. PCRs were performed using real-time quantitative PCR and normalized relative to GAPDH expression. Patient results represent individual patients analysed with independent PCR assays, each performed in triplicate, whereas normal donor results represent mean data from separate donors. Primer sequences are described previously (Follows et al., 2003a). Flow cytometry Flow cytometric analysis of diagnostic bone marrow samples was performed using multicolour flow cytometry and standard techniques at the Haematological Malignancy Diagnostic Service, Leeds, UK.


3651 In vivo DMS and UV footprinting In vivo DMS- and UV-footprinting experiments were performed as described previously (Kontaraki et al., 2000). With DMS footprinting, 1 mg of purified and piperidine-cleaved genomic DNA was subjected to LM-PCR amplification. With UV footprinting, 1 mg of purified DNA was used as starting material for TD-PCR amplification. Both LM- and TD-PCR techniques were performed as described previously (Kontaraki et al., 2000; Chen et al., 2001). CpG methylation studies Quantification of relative percentage CpG methylation was carried out as described (Tagoh et al., 2004). Briefly, genomic DNA was digested with methylation-sensitive restriction

enzymes (MaeII and HpaII), which digest at sites indicated in Figure 1b. Digestion was quantified using real-time PCR with primers designed to amplify across the digestion point using serial dilutions of nondigested genomic DNA of known concentration as a standard. The amount of DNA subjected to the restriction enzyme digestion was evaluated by the amplifications with control primers, which do not span the restriction enzyme recognition sites.

Acknowledgements GAF is a Leukaemia Research Fund Bennett Fellow, HT is funded by the Kay Kendall Leukaemia Fund. Research in CB’s laboratory is funded by the MRC, Wellcome Trust and Leukaemia Research Fund.

References Alcalay M, Meani N, Gelmetti V, Fantozzi A, Fagioli M, Orleth A, Riganelli D, Sebastiani C, Cappelli E, Casciari C, Sciurpi MT, Mariano AR, Minardi SP, Luzi L, Muller H, Di Fiore PP, Frosina G and Pelicci PG. (2003). J. Clin. Invest., 112, 1751–1761. Boultwood J, Rack K, Kelly S, Madden J, Sakaguchi AY, Wang LM, Oscier DG, Buckle VJ and Wainscoat JS. (1991). Proc. Natl. Acad. Sci. USA, 88, 6176–6180. Casas S, Nagy B, Elonen E, Aventin A, Larramendy ML, Sierra J, Ruutu T and Knuutila S. (2003). Leukemia Lymphoma, 44, 1935–1941. Chen HH, Kontaraki J, Bonifer C and Riggs AD. (2001). Sci. STKE, 2001, PL1. Dahl R, Walsh JC, Lancki D, Laslo P, Iyer SR, Singh H and Simon MC. (2003). Nat. Immunol., 10, 1029–1036. Dai XM, Ryan GR, Hapel AJ, Dominguez MG, Russell RG, Kapp S, Sylvestre V and Stanley ER. (2002). Blood, 99, 111–120. Debernardi S, Lillington DM, Chaplin T, Tomlinson S, Amess J, Rohatiner A, Lister TA and Young BD. (2003). Genes Chromosomes Cancer, 37, 149–158. Di Croce L, Raker VA, Corsaro M, Fazi F, Fanelli M, Faretta M, Fuks F, Lo Coco F, Kouzarides T, Nervi C, Minucci S and Pelicci PG. (2002). Science, 295, 1079–1082. Downing JR. (1999). Br. J. Haematol., 106, 296–308. Follows GA, Tagoh H, Lefevre P, Hodge D, Morgan GJ and Bonifer C. (2003a). EMBO J., 22, 2798–2809. Follows GA, Tagoh H, Lefevre P, Morgan GJ and Bonifer C. (2003b). Nucleic Acids Res., 31, 5805–5816. Gisselbrecht S, Fichelson S, Sola B, Bordereaux D, Hampe A, Andre C, Galibert F and Tambourin P. (1987). Nature, 329, 259–261.

Heidenreich O, Krauter J, Riehle H, Hadwiger P, John M, Heil G, Vornlocher HP and Nordheim A. (2003). Blood, 101, 3157–3163. Himes SR, Tagoh H, Goonetilleke N, Sasmono T, Oceandy D, Clark R, Bonifer C and Hume DA. (2001). J. Leukocyte Biol., 70, 812–820. Kelly LM and Gilliland DG. (2002). Annu. Rev. Genomics Hum. Genet., 3, 179–198. Kontaraki J, Chen HH, Riggs A and Bonifer C. (2000). Genes Dev., 14, 2106–2122. Lichtenberger C, Zakeri S, Baier K, Willheim M, Holub M and Reinisch W. (1999). J. Immunol. Methods, 227, 75–84. McGlynn H, Kapelko K, Baker A, Burnett A and Padua RA. (1997). Leukemia Res., 21, 919–923. Morris JF, Hromas R and Rauscher III FJ. (1994). Mol. Cell. Biol., 14, 1786–1795. Rosenbauer F, Wagner K, Kutok JL, Iwasaki H, Le Beau MM, Okuno Y, Akashi K, Fiering S and Tenen DG. (2004). Nat. Genet, 36, 624–630. Rountree MR, Bachman KE, Herman JG and Baylin SB. (2001). Oncogene, 20, 3156–3165. Sariban E, Mitchell T and Kufe D. (1985). Nature, 316, 64–66. Sasmono RT, Oceandy D, Pollard JW, Tong W, Pavli P, Wainwright BJ, Ostrowski MC, Himes SR and Hume DA. (2003). Blood, 101, 1155–1163. Sherr CJ, Rettenmier CW, Sacca R, Roussel MF, Look AT and Stanley ER. (1985). Cell, 41, 665–676. Springall F, O’Mara S, Shounan Y, Todd A, Ford D and Iland H. (1993). Leukemia, 7, 978–985. Tagoh H, Himes R, Clarke D, Leenen PJ, Riggs AD, Hume D and Bonifer C. (2002). Genes Dev., 16, 1721–1737. Tagoh H, Schebesta A, Lefevre P, Wilson N, Hume D, Busslinger M and Bonifer C. (2004). EMBO J., 23, 4275–4285.

Oncogene