Letters to the Editor 1885 between the t(8;21) karyotype and MDR-1 expression,5,6 expressing cells and t(8;21) patients,7,9 a that acts as a suggesting that this fusion gene may promote tumour suppressor gene in other contexts. the expression of MDR-1. We therefore tested this hypothesis directly by expressing the RUNX1-RUNX1T1 fusion as a single abnormality in human haematopoietic cell subsets and per- Acknowledgements formed Affymetrix microarray analysis (as described pre- viously)2,7 to determine whether this fusion had any effect on We thank Amanda Gilkes and Megan Musson (Cardiff University) the transcription of MDR . Using this approach, we for their technical assistance in processing microarray samples. generated independent replicate sets of data from control and We are grateful to the MRC for access to patient sample material RUNX1-RUNX1T1-matched CD34 þ cultures as well as matched enrolled in the NCRI clinical trials. This work was supported by sets constituting granulocytic (CD14lo,CD36lo,CD15hi)and Leukaemia Research, UK. hi monocytic (CD14 ) unilineage populations (isolated from day 6 A Tonks1, L Pearn1, KI Mills1, AK Burnett1 and RL Darley1 cultures by immunomagnetic sorting). cRNA was prepared from 1Department of Haematology, School of Medicine, Cardiff each sample and hybridized to Affymetrix human 133A University, Cardiff, UK oligonucleotide arrays, which allowed the simultaneous analysis E-mail: [email protected] of six MDR family gene members. In each of these populations, the expression of MDR genes was not significantly different from References controls (Figure 1a–c). In addition, using our cohort of French– American–British (FAB)-M2 patients, there was little difference in 1 Peterson LF, Zhang DE. The 8;21 translocation in leukemogenesis. MDR between those individuals with a t(8;21) Oncogene 2004; 23: 4255–4262. and those without this abnormality (Figure 1d). We could 2 Tonks A, Tonks AJ, Pearn L, Pearce L, Hoy T, Couzens S et al. therefore find no evidence that RUNX1-RUNX1T1 expression Expression of AML1-ETO in human myelomonocytic cells selec- directly influences MDR gene expression as a single abnormality tively inhibits granulocytic differentiation and promotes their self- renewal. Leukemia 2004; 18: 1238–1245. or in t(8;21) patients. One alternative explanation for the 3 Grimwade D, Walker H, Oliver F, Wheatley K, Harrison C, aforementioned observations in AML patients is that other Harrison G et al. The importance of diagnostic cytogenetics on coexisting abnormalities may be influencing the expression of outcome in AML: analysis of 1612 patients entered into the MRC MDR, as suggested by Schaich et al.6 AML 10 trial. The Medical Research Council Adult and Children’s We next addressed the issue of whether the t(8;21) Leukaemia Working Parties. Blood 1998; 92: 2322–2333. abnormality directly influences the susceptibility to chemother- 4 Lutterbach B, Sun D, Schuetz J, Hiebert SW. The MYND motif is required for repression of basal transcription from the multidrug apeutic agents. We therefore assessed the sensitivity of normal resistance 1 promoter by the t(8;21) fusion protein. Mol Cell Biol human cells (expressing RUNX1-RUNX1T1 as a single abnorm- 1998; 18: 3604–3611. ality) to a number of drugs commonly used to treat AML 5 Pearson L, Leith CP, Duncan MH, Chen IM, McConnell T, Trinkaus (Daunorubicin, Cytarabine, Fludarabine, Idarubicin or Etopo- K et al. Multidrug resistance-1 (MDR1) expression and functional side) in comparison with matched controls. Remarkably, none dye/drug efflux is highly correlated with the t(8;21) chromosomal of these agents differentially affected the growth of RUNX1- translocation in pediatric . Leukemia 1996; 10: 1274–1282. RUNX1T1-transduced cells (Figure 2a–e). As treatment of AML 6 Schaich M, Harbich-Brutscher E, Pascheberg U, Mohr B, Soucek S, commonly involves multiple drugs, we also determined the Ehninger G et al. Association of specific cytogenetic aberrations with effect of combining two or more of these chemotherapeutic mdr1 gene expression in adult myeloid leukemia and its implication agents (using drug concentrations that resulted in 50% reduction in treatment outcome. Haematologica 2002; 87: 455–464. in cell growth as a single agent). Again, we observed little 7 Tonks A, Pearn L, Musson M, Gilkes A, Mills K, Burnett A et al. difference in the in vitro growth response of RUNX1-RUNX1T1- RUNX1-RUNXIT1 induces over-expression of gamma-catenin in human CD34(+) cells, increasing self renewal and impairing expressing cells compared to controls (Figure 2f). granulocytic differentiation. Blood 2005; 106: 841A. Taken together, these data suggest that expression of RUNX1- 8 Yuan Y, Zhou L, Miyamoto T, Iwasaki H, Harakawa N, RUNX1T1 itself has no effect on the intrinsic susceptibility to Hetherington CJ et al. AML1-ETO expression is directly cytotoxic chemicals. This raises the alternative hypothesis that involved in the development of acute myeloid leukemia in the RUNX1-RUNX1T1 moderates the influence of secondary presence of additional . Proc Natl Acad Sci USA 2001; 98: abnormalities, which are required for RUNX1-RUNX1T1- 10398–10403. 8 9 Muller-Tidow C, Steffen B, Cauvet T, Tickenbrock L, Ji P, Diederichs expressing cells to undergo leukaemic transformation. For S et al. Translocation products in acute myeloid leukemia activate example, it is known that RUNX1-RUNX1T1 specifically the Wnt signaling pathway in hematopoietic cells. Mol Cell Biol upregulates the expression of g-catenin in RUNX1-RUNX1T1- 2004; 24: 2890–2904.

Oncogenic kinases of myeloproliferative disorders induce both protein synthesis and G1 activators

Leukemia (2006) 20, 1885–1888. doi:10.1038/sj.leu.2404361; they may progress to an acute syndrome. V617F of the published online 10 August 2006 Janus kinase 2 (JAK2) kinase are found in polycythemia vera, essential thrombocythemia and idiopathic myelofibrosis. BCR- ABL fusion occurs in chronic myeloid leukemia, and a variety of Myeloproliferative disorders (MPDs) are clonal proliferative fusions involving platelet-derived growth factor and fibroblast diseases of the hematopoietic stem cells. After an initial phase, growth factor (FGF) receptors are found in other MPDs.1 Fusion

Leukemia Letters to the Editor 1886 Table 1 Selection of discriminant genes classified by Onto-Express

GO ID Biological process P-value Gene Symbol Probe set DS

GO:0006364 rRNA processing 3.07EÀ11 Nola1 1418305_s_at 2.10 Exosc2 1426630_at 1.96 Rnu3ip2 1451293_at 1.46 Mphosph10 1429080_at 1.42 Ddx56 1423815_at 1.41 Ebna1bp2 1428315_at 1.31 Exosc1 1452012_a_at 1.31

GO:0007046 Ribosome biogenesis 4.82EÀ11 Rpl12 1435655_at 1.73 1110017C15Rik 1448480_at 1.63 Rrs1 1456865_x_at 1.63 Nol5a 1426533_at 1.59 Gtpbp4 1450873_at 1.40 2610012O22Rik 1423823_at 1.30

GO:0006412 Protein biosynthesis 3.60EÀ05 Rpl41 1454639_x_at 1.90 Mrps18b 1451164_a_at 1.82 Rpl3 1438527_at 1.79 Rps9 1426958_at 1.69 Eif3s1 1426394_at 1.53 Nola2 1416605_at 1.51 Itgb4bp 1427578_a_at 1.47 Eef1e1 1449044_at 1.45 Eif3s4 1417718_at 1.38

GO:0006457 Protein folding 6.72EÀ03 Dnajc11 1433880_at 1.38 Cct3 1448178_a_at 1.33 Ero1l 1419030_at 1.32 Hyou1 1423291_s_at 1.31

GO:0006511 Ubiquitin-dependent protein catabolism 1,47EÀ02 Usp39 1437007_x_at 1.51 Siah2 1448171_at 1.30 Usp10 1448230_at 1.29

GO:0000398 Nuclear mRNA splicing, via spliceosome 1,65EÀ02 Mki67ip 1424001_at 1.88 Pprc1 1426381_at 1.86 Sfrs2 1415807_s_at 1.46

GO:0000074 Regulation of cell cycle 1.65EÀ02 Axl 1423586_at 1.65 Ccnd2 1430127_a_at 1.51 Cdc25a 1417132_at 1.43 Abbreviation: DS, discriminating score; rRNA, ribosomal RNA.

kinases that result from a translocation are made of a starved for 7 h in RPMI plus 0.5% FCS. RNA integrity was constitutively activated kinase domain and an N-terminal region controlled by microanalysis (Agilent Bioanalyzer, Palo Alto, CA, encoded by a partner gene. The activated kinase triggers USA). Preparation of cRNA, hybridizations, washes, detection sustained proliferation and survival of the hematopoietic and quantification were performed as recommended by the cells, but maturation is not affected. supplier (www.Affymetrix.com). Data were analyzed by the We studied the gene expression profiles of murine Ba/F3 cells robust multichip average (RMA) method in R using Bioconduc- transfected by various oncogenic MPD kinases by using whole- tor and associated package. Before analysis, a filtering process genome Affymetrix 430 2.0 mouse oligonucleotide microarrays removed from the data set the genes with low and poorly (www.Affymetrix.com). Ba/F3 cells were grown in Rosewell measured expression as defined by an expression value inferior Park Memorial Institute (RPMI) medium supplemented with to 100 U in all samples, retaining 17.885 genes/expressed 10% fetal calf serum (FCS) and interleukin-3 (IL-3). Expression of sequence tags (ESTs). For paired samples, RNA was prepared an MPD kinase bypasses the IL-3 dependence of Ba/F3 cells. independently from different cultures of cells. The correlation RNA was extracted by using Trizol (Trizol Reagent, Invitrogen between paired samples ranged between 0.97 and 0.98. Life Technologies, Carlsbad, CA, USA) from frozen pellets of: (i) Gene expression profiles of Ba/F3 cells transfected by fusion Ba/F3 cells; (ii) Ba/F3 transfected with different pCDNA express- or mutated kinases (nine samples: BCR-ABL, two BCR-FGFR1, ion vectors expressing a mutant, kinase-defective FOP-FGFR1 two CEP1-FGFR1, two FOP-FGFR1, two V617F JAK2) were KD or an oncogenic MPD kinase (BCR-ABL, FOP-FGFR1, CEP1- compared to that of control cells (six samples), including FGFR1, BCR-FGFR1); and (iii) Ba/F3 transfected by the MIGR parental Ba/F3 cells (four samples) and Ba/F3 expressing a vector, either empty (MIGR) or with JAK2 (JAK2 WT, mutated kinase-defective mutant of FOP-FGFR1 (two samples). Super- V617F JAK2 or mutated V617F IND JAK2, that is, able to grow vised analysis, based on 17 885 filtered probe sets (Supplemen- independently of IL-3).2 Before RNA extraction, cells were tary Table 1), identified 294 differentially expressed probe sets

Leukemia Letters to the Editor 1887

Figure 1 Gene expression profiling of Ba/F3 cell samples identifies a molecular signature of activated MPD kinases. (a) Hierarchical clustering of 15 samples using the 294 genes identified as discriminator between Ba/F3 cells transfected by activated kinases (nine samples: BCR-ABL, two BCR-FGFR1, two CEP1-FGFR1, two FOP-FGFR1 and two V617F JAK2) to that of parental Ba/F3 cells (four samples) and Ba/F3 expressing a kinase- defective mutant of FOP-FGFR1 (two samples). Each row represents a gene and each column a sample. The log2-transformed expression level of each gene in a single sample is relative to its median abundance across all samples and is depicted according to the color scale shown at the bottom. Red and green indicate expression levels, respectively, above and below the median. The magnitude of deviation from the median is represented by the color saturation. The dendrogram of samples (above matrix) represents overall similarities in gene expression profiles and is zoomed in panel b. Branches of the dendrograms are color-coded as follows: red for fusion kinase-expressing Ba/F3 cells and green for control cells. Some genes included in the signature (framed in red for the upregulated genes and in green for the downregulated genes) are noted to the right of the data matrix and referenced by their abbreviation as used in EntrezGene. A discriminating score (DS) was calculated for each gene. DS ¼ (M1ÀM2)/(S1 þ S2) where M1 and S1, respectively, represent mean and s.d. of expression levels of the gene in subgroup 1, and M2 and S2 in subgroup 2. Confidence levels were estimated by 100 random permutations of samples. (a) ‘Leave-one-out’ procedure estimated the accuracy of prediction of the signatures and the validity of our supervised analysis. (b) (Top) Dendrogram of samples. (Down) Correlation between the molecular grouping based on the combined expression of the 294 genes and the status of samples. (c) Western blot analysis of cyclin D2 in Ba/F3 expressing activated MPD kinases. NP-40-extracted were separated by gel electrophoresis (sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE)), transferred onto membrane (Hybond-C Extra, GE Healthcare UK, Buckinghamshire, UK) and probed with rabbit polyclonal anti-cyclin D2 (M-20, Santa Cruz Biotechnology, Santa Cruz, CA, USA). Cyclin D2 (top) is upregulated in Ba/F3 cells expressing MPD kinases compared to control Ba/F3 cells (untransfected, FOP-FGFR1 KD, MIGR and JAK2 WT). Total cell lysates were probed with mouse monoclonal anti-a-tubulin (B-5-1-2, Sigma-Aldrich, Saint-Quentin Fallavier, France) to compare the amount of protein in the lysates (bottom).

(theoretical number of produced false positives ¼ 1.7) (Figure 1a PWP1) proteins, eucaryotic initiation and elongation factors and b), representing 228 genes and eight ESTs, of which 188 (EIF1A, EIF3S1, EIF3S4, EEF1E1), and components of the were upregulated and 48 downregulated in activated kinase- exosome (EXOSC1, EXOSC2, EXOSC6). Upregulated genes also expressing cells (Supplementary Tables 2 and 3). encode proteins of the NOL5A-associated pre-ribosomal ribo- To translate the RNA expression profiles into functionality, complex involved in pre-rRNA processing discriminator genes/ESTs were interrogated by Onto-Express. (NOL5A, PPAN, NOLC1 and BXDC2). The gene encoding Table 1 represents the most significant (P-value inferior at EBNA1BP2 was upregulated; it encodes a protein that binds to 3 Â 10À2) and most often represented (including at least three nucleolar FGF34 and is regularly upregulated in tumors. The genes) biological processes. Many of the upregulated genes most upregulated sequence was GAS5, a non-protein-coding encode nucleolar proteins involved in ‘ribosome biogenesis’ multiple small nucleolar RNA. (GO:0007046; six genes, P ¼ 4.28 Â 10À11), ‘rRNA processing’ Other significant processes included ‘protein folding’ (GO: (GO:0006364; seven genes, P ¼ 3.07 Â 10À11) and ‘protein 0006457; four genes, P ¼ 6.72 Â 10À03), ‘ubiquitin-dependent pro- biosynthesis’ (GO:0006412; nine genes, P ¼ 3.60 Â 10À05). tein catabolism’ (GO:0006511; three genes, P ¼ 1.47 Â 10À02), Upregulated genes encode nucleolar proteins (CIRH1A, LARP1, ‘nuclear mRNA splicing, via spliceosome’ (GO:0000398; three NOL1, NOL11, NOL5, NOL5A, NOLA1, NOLA2, NOLC1, genes, P ¼ 1.65 Â 10À02) and ‘regulation of cell cycle’ (GO: MKI67IP, SFRS2, SURF6),3 ribosomal proteins (RPL3, RPL12, 0000074; three genes, P ¼ 1.65 Â 10À02). The second major RPL41, RPS9, RRS1), small nuclear ribonucleoproteins and category of upregulated genes encodes CCND2 (cyclin D2) and interactors (U3/MPHOSPH10, LSM2, RNU22, RNU3IP2), com- CDC25A, two major regulators needed for G1 progression. ponents of RNA polymerase I (POLR1A, POLR1B), II (POLR2H, CCND2 RNA was found upregulated by BCR-ABL in previous TAF9) and III (POLR3E, POLR3H), DEAD-box (DDX18, DDX56) gene expression studies.5–7 Cyclin D2 is necessary for BCR-ABL- and WD repeat (WDR4, WDR43, WDR74, WDR77, GRWD1, induced activity.7 Inhibition of V617F JAK2 correlates with

Leukemia Letters to the Editor 1888 decreased expression of cyclin D2.8 Other G1 cyclins may play Acknowledgements a role in the oncogenic activity of fusion kinases, but cyclin D2 seems to be a rate-limiting element. We used Western blot We thank B Delaval and MJ Mozziconacci for discussions and analysis to validate the differential expression of cyclin D2. The advices, and P Dubreuil (Marseille) and NC Cross (Salisbury, UK) amount of cyclin D2 protein was increased in Ba/F3 cells for reagents. This work was supported by the Institut Paoli- expressing activated kinases as compared to controls (Figure 1c), Calmettes, Inserm, and grants from the Institut National du in agreement with mRNA expression results. Cancer/Cance´ropoˆle (ACI2005) and the Association Laurette directly or indirectly regulates the G1 phase of the cell Fugain (2005).

cycle. The list of upregulated genes included MYC. Many genes 1,3 1,3 1 2 upregulated by MYC and NMYC oncogenes were also H Lelie`vre , N Cervera , P Finetti , F Delhommeau , W Vainchenker2, F Bertucci1 and D Birnbaum1 upregulated in our experiments, including CCND2, CDC25A 1 Laboratoire d’Oncologie Mole´culaire, Centre de Recherche and others (DDX18, EBNA1BP2, EEF1E1, MAT2A, MKI67IP, en Cance´rologie de Marseille, UMR599 Inserm et Institut NOL5A, NOLA1, PHB, SFRS2, SHMT1, SLC16A1, SURF6, SRM, Paoli-Calmettes, Marseille, France and 9 RPL3, RPL12, RPL41, RPS9 and RRS1). This similarity suggests 2U362 Inserm, Institut Gustave Roussy, Villejuif, France that MYC proteins and MPD kinases have similar oncogenic E-mail: [email protected] effects, whose main target would be the CDKN2-RB protein 3These authors contributed equally to this work. pathway during the G1 phase of the cell cycle. Once induced, MYC may in turn act on the transcription of G1/S regulators and genes involved in protein synthesis. MPD fusion kinases are thought to target the hematopoietic stem cell. Activation of MYC is in perfect agreement with what we know of stem cell References proliferation. A similar program was also turned on by IL-3 stimulation (not shown). 1 De Keersmaecker K, Cools J. Chronic myeloproliferative disorders: a tyrosine kinase tale. Leukemia 2006; 20: 200–205. Downregulated genes were more difficult to classify with 2 James C, Ugo V, Le Couedic JP, Staerk J, Delhommeau F, Lacout C Onto-Express, but several encode proteins with known or et al. A unique clonal JAK2 mutation leading to constitutive signalling potential inhibitory function such as protein inhibitor of causes polycythaemia vera. Nature 2005; 434: 1144–1148. activated STAT3 (PIAS3), an inhibitor of signal transducer and 3 Lamond A, Mann M. Nucleolar Proteome Database v 2.0 2005. activator of transcription 3 (STAT3), one of the main substrates of [cited Sep 06]; available from: www.lamondlab.com/NoPDB. MPD kinases, and regulator of CDC25A,10 Erbin and PLZF/ 4 Reimers K, Antoine M, Zapatka M, Blecken V, Dickson C, Kiefer P. ZBTB16, a MYC repressor.11 NoBP, a nuclear fibroblast growth factor 3 binding protein, is cell cycle regulated and promotes cell growth. Mol Cell Biol 2001; 21: We tested the validity of our classification by the ‘leave-one- 4996–5007. out’ cross-validation method. Iteratively, one of the 15 samples 5 Parada Y, Banerji L, Glassford J, Lea NC, Collado M, Rivas C et al. was removed, and a multigene predictor was generated from BCR-ABL and interleukin 3 promote haematopoietic cell prolifera- the remaining samples: 93% of samples were correctly tion and survival through modulation of cyclin D2 and p27Kip1 assigned by the predictors with a sensitivity of 89% and a expression. J Biol Chem 2001; 276: 23572–23580. specificity of 100%. 6 Deininger MW, Vieira S, Mendiola R, Schultheis B, Goldman JM, Melo JV. BCR-ABL tyrosine kinase activity regulates the expression Thus, in Ba/F3 cells, MPD fusion kinases induce both G1 of multiple genes implicated in the pathogenesis of chronic activators and protein synthesis components, thus starting the myeloid leukemia. Cancer Res 2000; 60: 2049–2055. cell proliferation machinery. This effect may be mediated by the 7 Jena N, Deng M, Sicinska E, Sicinski P, Daley GQ. Critical role for PI3 kinase (PI3K)–AKT–TOR pathway, which controls and cyclin D2 in BCR/ABL-induced proliferation of hematopoietic coordinates both protein synthesis and early phases of the cell cells. Cancer Res 2002; 62: 535–541. cycle. Prominent downstream targets of the AKT pathway are 8 Walz C, Crowley BJ, Hudon HE, Gramlich JL, Neuberg DS, Podar 12 K et al. Activated JAK2 with the V617F point mutation promotes cyclins D1, D2 and MYC. These conclusions have to be G1/S-phase transition. J Biol Chem 2006; 281: 18177–18183. confirmed by repeating the experiments in other cell lines and 9 Zeller KI, Jegga AG, Aronow BJ, O’Donnell KA, Dang CV. An by profiling samples of various human MPDs. This work has two integrated database of genes responsive to the Myc oncogenic general implications. First, it may help define the initial steps of transcription factor: identification of direct genomic targets. oncogenesis. Cell quiescence may be defined by the absence Genome Biol 2003; 4: R69. of protein synthesis, whereas proliferation starts with nucleolar 10 Barre B, Vigneron A, Coqueret O. The STAT3 transcription factor is a target for the Myc and riboblastoma proteins on the Cdc25A activity, ribosome biogenesis and rRNA processing. Second, promoter. J Biol Chem 2005; 280: 15673–15681. MPDs could benefit from treatment targeting not only the 11 McConnell MJ, Chevallier N, Berkofsky-Fessler W, Giltnane JM, activated kinases but also the PI3K–AKT–TOR signalling Malani RB, Staudt LM et al. Growth suppression by acute pathway and the G1 phase of the cell cycle, in synergy promyelocytic leukemia-associated protein PLZF is mediated with anti-kinase drugs and/or in case of resistance. G1 targeting by repression of c-myc expression. Mol Cell Biol 2003; 23: is frequent in many types of cancer. The signature defined 9375–9388. 12 Gera JF, Mellinghoff IK, Shi Y, Rettig MB, Tran C, Hsu JH et al. AKT here can be used in transcriptome studies of any type of activity determines sensitivity to mammalian target of rapamycin cancer to identify G1 activation, classify tumors and use the (mTOR) inhibitors by regulating cyclin D1 and c-myc expression. appropriate drugs. J Biol Chem 2004; 279: 2737–2746.

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

Leukemia