Leukemia (2007) 21, 494–504 & 2007 Nature Publishing Group All rights reserved 0887-6924/07 $30.00 www.nature.com/leu ORIGINAL ARTICLE

Molecular signature of CD34 þ hematopoietic stem and progenitor cells of patients with CML in chronic phase

E Diaz-Blanco1,6, I Bruns1,6, F Neumann1, JC Fischer2, T Graef1, M Rosskopf3, B Brors4, S Pechtel1, S Bork1, A Koch1, A Baer1, U-P Rohr1, G Kobbe1, A von Haeseler5, N Gattermann1, R Haas1 and R Kronenwett1,7

1Department of Hematology, Oncology and Clinical Immunology, University of Duesseldorf, Duesseldorf, Germany; 2Institute of Transplantation Diagnostics and Cell Therapeutics, University of Duesseldorf, Duesseldorf, Germany; 3Institute of Bioinformatics, University of Duesseldorf, Duesseldorf, Germany; 4Theoretical Bioinformatics, German Cancer Research Center, Heidelberg, Germany and 5Max F. Perutz Laboratories, Center for Integrative Bioinformatics; University of Vienna; Medical University of Vienna; University of Veterinary Medicine Vienna, Vienna, Austria

In this study, we provide a molecular signature of highly BCR-ABL oncoprotein.3 It constitutively activates mitogenic enriched CD34 þ cells from bone marrow of untreated patients signaling pathways such as rat sarcoma viral oncogene homolog with chronic myelogenous leukemia (CML) in chronic phase in (RAS)/mitogen-activated kinase (MAPK) pathways, the comparison with normal CD34 þ cells using microarrays covering 8746 . Expression data reflected several BCR- janus kinase (JAK)/signal transducer and activator of transcrip- ABL-induced effects in primary CML progenitors, such as tion (STAT) pathway, phosphoinositide-3 (PI3) kinase pathway transcriptional activation of the classical mitogen-activated and the MYC pathway. In addition, BCR-ABL induces adhesive protein kinase pathway and the phosphoinositide-3 kinase/AKT and cytoskeletal abnormalities resulting in deregulation of pathway as well as downregulation of the proapoptotic proliferation, apoptosis and release of progenitors from bone IRF8. Moreover, novel transcriptional changes in comparison marrow (BM).4,5 Several studies suggest that the expansion of with normal CD34 þ cells were identified. These include upregulation of genes involved in the transforming growth malignant cells in CML is not directly driven by the neoplastic 6 factorb pathway, fetal hemoglobin genes, , stem cell but by lineage-committed progenitor cells. Interest- sorcin, tissue inhibitor of metalloproteinase 1, the neuroepithe- ingly, treatment of patients with the tyrosine kinase inhibitor lial cell transforming gene 1 and downregulation of selenopro- imatinib results in the persistence of BCR-ABL-positive cells in tein P. Additionally, genes associated with early hematopoietic most of the patients in complete cytogenetic response, suggest- stem cells (HSC) and leukemogenesis such as HoxA9 and ing that blocking of BCR-ABL induces an inhibition of progenitor MEIS1 were transcriptionally activated. Differential expression 7 of differentiation-associated genes suggested an altered com- cells, whereas primitive leukemic stem cells are spared. position of the CD34 þ cell population in CML. This was A limitation of the existing data about BCR-ABL-mediated confirmed by subset analyses of chronic phase CML CD34 þ effects is that they are predominantly derived from human and cells showing an increase of the proportion of megakaryocyte- murine BCR-ABL-transformed cell lines and mouse models,3 erythroid progenitors, whereas the proportion of HSC and which might not be representative for chronic phase CML. The granulocyte–macrophage progenitors was decreased in CML. In conclusion, our results give novel insights into the biology of cell population for studying the pathophysiology of CML are CML and could provide the basis for identification of new certainly primary hematopoietic stem and progenitor cells from therapeutic targets. patients with CML. Leukemia (2007) 21, 494–504. doi:10.1038/sj.leu.2404549; The assessment of primary CML cells by means of gene published online 25 January 2007 expression analyses comparing BCR-ABL-positive cells with Keywords: CML; hematopoietic stem cells; CD34 þ cell subsets; normal cells,8–10 resulted in expression profiles associated with profiling; BCR-ABL signaling response to therapy11–15 or progression to advanced phases of disease.16,17 In a previous gene expression study using an array covering 1185 genes, we could confirm some previously Introduction described functional effects of BCR-ABL in highly enriched CD34 þ hematopoietic stem and progenitor cells.8 Although Chronic myelogenous leukemia (CML) is a malignant disorder our study gave novel insights into BCR-ABL-induced effects in of the hematopoietic stem cell (HSC) characterized by primary CML cells, no study exists to date, which provides a a reciprocal translocation between 9 and 22 large-scale expression analysis of chronic phase CML CD34 þ (t(9;22)(q34;q11)).1 The translocation results in formation of the stem and progenitor cells in comparison with the normal BCR-ABL fusion oncogene encoding a protein with constitutive counterparts. tyrosine kinase activation, which plays a central role in Therefore, we examined highly enriched CD34 þ cells from the pathogenesis of the disease.2 Numerous mechanisms are BM of untreated patients with CML in first chronic phase using involved in the malignant transformation orchestrated by the microarrays covering 8746 genes and refined the molecular signature of hematopoietic stem and progenitor cells in CML. Correspondence: Dr R Kronenwett, Department of Hematology, Oncology and Clinical Immunology; University of Du¨sseldorf, Moorenstr. 5, 40225 Du¨sseldorf, Germany. E-mail: [email protected] Materials and methods 6These authors contributed equally to this work. 7Current address: Bayer HealthCare AG, Diagnostics Research, Leverkusen, Germany. Patients and cells Received 1 November 2006; accepted 16 November 2006; published After informed consent BM mononuclear cells were obtained by online 25 January 2007 density centrifugation as previously described18 from eight Molecular signature of CML CD34 þ cells D-B Elena et al 495 healthy volunteers and from nine patients with untreated newly Optical 96-well Reaction Plate (Applera) according to the diagnosed Ph þ CML in chronic phase for gene expression instructions of the manufacturer using commercially available analyses as well as from three healthy volunteers and from three assays-on-demand. GAPDH mRNA served as external control patients for subset analyses (patients characteristics: Supple- for relative quantification. The following Assay-on-Demand mentary Table 1). From three of the nine patients (patient Gene Expression Products (Applera) were used: GATA1 numbers 1–3), we also obtained peripheral blood (PB) mono- (Hs00231112_m1), leptin receptor (LEPR) (Hs00174497_m1), nuclear cells. Samples were processed within 2 h following TAL1 (Hs00268434_m1), tissue inhibitor of metalloproteinase 1 puncture. CD34 þ cells were positively selected using two (TIMP1) (Hs00171558_m1), ACSL (acyl-CoA synthetase long- rounds of midiMACS immunomagnetic separation (Miltenyi chain) 5 (Hs00212106_m1), CRHBP (Hs00181810_m1), seleno- Biotec, Bergisch Gladbach, Germany) following Ficoll density protein P (SEPP1) (Hs00193657_m1), ELA2 (Hs00236952_m1) centrifugation as described previously.18 Purities of CD34 þ and GAPDH (Hs99999905_m1). CT values were calculated by cells for expression analyses as assessed by flow cytometry the ABI PRISM software and relative gene expression levels were varied between 98.9 and 99.9%. expressed as the difference of CT values of target gene and GAPDH (DCT). Given a PCR efficacy of 2 a DCT value of 1 reflects a fold-change of 2. RNA isolation, cRNA labeling and hybridization to microarrays Total RNA was isolated from CD34 þ cells using the RNeasy Immunofluorescence staining and flow cytometry Mini-Kit (Qiagen, Hilden, Germany) according to the manu- Flow cytometric analysis of CD34 þ subsets (stem cells and facturer’s instructions. The amount of extracted RNA was myeloid progenitors) was performed as described previously.22 quantified using the NanoDrop spectrophotometer (NanoDrop In brief, following immunomagnetical selection, CD34 þ cells Technologies, Welmongton, DE, USA). Total RNA (median: were stained with fluorescein isothiocyanate-conjugated anti- 7.0 mg; range: 0.7–20 mg) was used to generate biotin-labeled bodies specific for lineage markers CD3 (clone S4.1), CD4 cRNA (median: 25 mg; range: 2.1–68 mg) by means of Enzo (S3.5), CD7 (CD7-6B7), CD8 (385), CD10 (5-1B4), CD14 BioArray HighYield RNA Transcript Labeling Kit (Affymetrix Ltd, (TUK4), CD19 (SJ25-C1) (Caltag, South San Francisco, CA, UK). Following fragmentation, labelled cRNA was hybridized to USA), CD2 (RPA-2.10), CD11b (ICRF44), CD20 (2H7), CD56 Affymetrix HG-Focus GeneChips stained according to the (B159), GPA (GA-R2) (Becton Dickinson-PharMingen, San manufacturer’s instructions. The quality of extracted RNA and Diego, CA, USA), and ECD-conjugated anti-CD45RA (MEM56) of cRNA was controlled using an Agilent Bioanalyzer 2100 (Beckman Coulter, Miami, FL, USA), phycoerythrin-conjugated (Agilent Technologies, Waldbronn, Germany). anti-interleukin (IL)-3Ralpha (9F5) (Becton Dickinson-Phar- Mingen), allophycocyanin conjugated anti-CD38 (HIT2) (Caltag) and PerCP-conjugated anti-CD34 (HPCA-2) (Becton Quantification, normalization and statistical analysis Dickinson-PharMingen). Dead cells were excluded by propi- For quality control, normalization and data analysis, we utilized dium iodide staining. Isotype-matched control monoclonal the affy package of functions of statistical scripting language antibodies were used to determine the level of background ‘R’ integrated into the Bioconductor project (http://www. staining. Cells were analyzed and sorted using a double laser bioconductor.org/). Using histograms of perfect match intensi- (488 nm/350 nm Enterprise II þ 647 Spectrum) high-speed cell ties, 5’–3’ RNA degradation side-by-side plots, or scatter plots, sorter (Epics ELITE ESP, BeckmanCoulter, Miami, FL, USA). we estimated the quality of probes and hybridizations. Only samples were included into the analysis, which had a glyceraldehyde-3-phosphate dehydrogenase (GAPDH) 30/50 Statistical analysis ratio below 2. To normalize raw data, we used a method of For calculation of statistically significant differences between variance stabilizing transformations.19 For identification of two groups in quantitative real-time RT-PCR and subset analysis differentially expressed genes, we used the significance analysis the t-test for unpaired samples was used. A Po0.05 was of microarrays (SAM) algorithm v2.23 (http://www-stat.stanfor- considered to be statistically significant. d.edu/_tibs/SAM/), which contains a sliding scale for false discovery rate (FDR) of significantly up- and downregulated genes.20 All data were permuted 1000 times by using the two Results and discussion classes, unpaired data mode of the algorithm. As cutoff for significance an estimated FDR of 0.1% (comparison CML vs Gene expression signature of CD34 þ cells from normal) or 5% (comparison CML BM vs CML PB) was chosen by patients with CML in chronic phase the tuning parameter delta of the software. The significance Immunomagnetically enriched CD34 þ hematopoietic stem level of each gene was given by the Q-value (lowest FDR at and progenitor cells from BM of nine patients with untreated which the gene is called significant). Moreover, a cutoff for fold- BCR-ABL-positive CML in chronic phase and from eight healthy change of differential expression of 1.2 was used. volunteers were examined using Affymetrix HG Focus arrays covering 8746 genes. The complete results of our array experiments are available in the gene expression omnibus Quantitative real-time reverse transcription-polymerase database (www.ncbi.nlm.nih.gov/geo/; accession no.: chain reaction GSE5550) according to MIAME standards.23 Our gene expres- Corroboration of RNA expression data was performed by real- sion analyses fulfilled the recently published consensus guide- time reverse transcription-polymerase chain reaction (RT-PCR) lines from three European leukemia networks.24 using the ABI PRISM 7900HT Sequence Detection System Comparing malignant BM CD34 þ cells from patients with Instrument (Applied Biosystems, Applera Deutschland GmbH, CML with normal BM CD34 þ cells 1539 genes were signi- Darmstadt, Germany). Total RNA was reverse transcribed as ficantly differentially expressed (fold change: X1.2 or p0.83; described previously.21 PCR was performed in a MicroAmp Q-value: o0.1%) (Supplementary Figure 1a; Supplementary

Leukemia Molecular signature of CML CD34 þ cells D-B Elena et al 496 Table 2) indicating a distinct molecular phenotype of chronic In the following, we describe the expression data comparing phase CML CD34 þ cells. A total of 1420 genes had a higher normal CD34 þ cells with CML CD34 þ cells by allocating and 119 genes a lower expression in CML. We then compared genes into functional groups (Table 1). the expression profiles of sedentary leukemic CD34 þ cells from BM with those of circulating CD34 þ cells from PB of untreated patients with CML and found no differentially expressed genes CML CD34 þ cells showed a higher expression level of with sufficient significance (Q-value: o5%) (Supplementary genes involved in BCR-ABL-signaling Figure 1b). This is in line with our previous findings made with a First, we were interested in genes encoding substrates of BCR- smaller number of 1185 genes.8 Interestingly, it is different to ABL.3 We found an upregulation of the BCR-ABL adapter normal CD34 þ cells, as we and others had observed differential Grb2 and Crkl, of the cytoskeletal protein talin as well expression of genes when comparing normal CD34 þ cells from as of the signaling molecules STAT3, STAT4, STAT5, VAV3 and BM with those from PB.18,25 the Ras GTPase-activating protein G3BP2 (Table 1; Figure 1).

Table 1 Selection of genes with significantly differential gene expression in CD34+ cells from patients with CML in comparison with those from healthy volunteers

Gene symbol Gene title Fold change (CML/normal)

BCR-ABL signaling AKT1 v-akt murine thymoma viral oncogene homolog 1 1.36 BCL2L1 BCL2-like 1 (BCL-XL) 1.35 CRKL v-crk sarcoma virus CT10 oncogene homolog (avian)-like 1.42 G3BP2 Ras-GTPase activating protein SH3 domain-binding protein 2 1.39 GRB2 Growth factor receptor-bound protein 2 1.20 HNRPK Heterogeneous nuclear ribonucleoprotein K 1.20 KRAS v-Ki-ras2 Kirsten rat sarcoma viral oncogene homolog 1.44 LYN V-yes-1 Yamaguchi sarcoma viral related oncogene homolog 1.56 MAP2K1 Mitogen-activated protein kinase kinase 1 (MEK1) 1.21 MAP4K5 Mitogen-activated protein kinase kinase kinase kinase 5 (GCKR) 1.31 MAPK1 Mitogen-activated protein kinase 1 (ERK2) 1.42 MYC v-myc myelocytomatosis viral oncogene homolog 1.85 NFKB1 NF kappa B 1 (p105) 1.21 NRAS Neuroblastoma RAS viral (v-ras) oncogene homolog 1.26 PIK3C3 Phosphoinositide-3-kinase, class 3 1.28 PIK3CA Phosphoinositide-3-kinase, catalytic, alpha polypeptide 1.20 PIK3CB Phosphoinositide-3-kinase, catalytic, beta polypeptide 1.35 RAC1 Ras-related C3 botulinum toxin substrate 1 1.27 RAC2 Ras-related C3 botulinum toxin substrate 2 1.41 STAT3 Signal transducer and activator of transcription 3 1.26 STAT4 Signal transducer and activator of transcription 4 1.32 STAT5A Signal transducer and activator of transcription 5A 1.42 VAV3 vav 3 oncogene 1.25 DUSP1 dual specificity phosphatase 1 0.51

Proliferation and apoptosis CASP3,6,10 Caspases 3,6,10 1.31–2.30 CCNG1,H Cyclins G1, H 1.36–1.72 CDC Cell division cycle proteins 16,23,27,37 1.30–1.32 CDK2,7,8 cyclin-dependent kinases 2,7,8 1.34–1.59 CDK2AP1 CDK2-associated protein 1 1.45 FAS FAS 1.26 MCM3 Minichromosome maintenance deficient 3 1.38 ORC2,4,5L Origin recognition complex, subunit 2-like (4-like, 5-like) 1.26–1.50 POLE2 DNA polymerase, epsilon 2 (p59 subunit) 1.47 POLE3 DNA polymerase, epsilon 3 (p17 subunit) 1.26 TFDP1 transcription factor Dp-1 1.37 IRF8 binding protein 8 0.23

Cytoskeleton, adhesion and stem cell niche CADH1 E-cadherin 1.94 CTNNA1 Catenin, alpha 1 1.33 FLNA Filamin A, alpha 1.37 ICAM4 Intercellular adhesion molecule 4 1.88 ITGAE,2B Integrin, alpha E/alpha 2b 1.31–2.74 PARVB Parvin, beta 1.71 TEK Tyrosine-protein kinase receptor Tie-2 1.34 TLN1 Talin 1.38 VCL Vinculin 1.46 ACTG1 Actin, gamma 1 0.71 CADH2 N-cadherin 0.72 FLNB filamin B, beta 0.69 SELL L- 0.41

Leukemia Molecular signature of CML CD34 þ cells D-B Elena et al 497 Table 1 (Continued).

Gene symbol Gene title Fold change (CML/normal)

Growth factors and their receptors CSF1 Macrophage colony stimulating factor 1.41 CSF2RB Granulocyte-macrophage colony-stimulating factor receptor 2.35 IL1B interleukin 1, beta 1.63 IL7 interleukin 7 1.35 NET1 Neuroepithelial cell transforming gene 1 2.66 SMAD2,4 Mothers against DPP homolog 2,4 1.35–1.55 TGFB1 transforming growth factor, beta 1 1.94 FGFR1 Fibroblast growth factor receptor 1 0.75 FLT3 fms-related tyrosine kinase 3 0.44 IL7R interleukin 7 receptor 0.69

Self-renewal and early stem cells ALDH1A1 Aldehyde dehydrogenase 1 family, member A1 3.28 GATA2 GATA binding protein 2 1.85 HOXA10 Homeo box A10 1.36 HOXA9 Homeo box A9 1.48 LMO2 LIM domain only 2 (rhombotin-like 1) 1.38 MEIS1 Myeloid ecotropic viral integration site 1 homolog 1.44 PBX3 Pre-B-cell leukemia transcription factor 3 1.79 PCGF4 polycomb group ring finger 4 1.38 TAL1 T-cell acute lymphocytic leukemia 1 1.95 ZNFN1A1 Zinc-finger protein, subfamily 1A, 1 (Ikaros) 1.28

Differentiation EPOR 1.22 GATA1 GATA binding protein 1 1.99 HBB Hemoglobin, beta 2.87 HBD Hemoglobin, delta 4.06 HBG1/G2 Hemoglobin, gamma A/gamma G 3.94 KLF1 Kruppel-like factor 1 (erythroid) 2.32 TFR2 2 2.54 BCL6 B-cell CLL/lymphoma 6 0.59 BLNK B-cell linker 0.29 CEBPD CCAAT/enhancer binding protein (C/EBP), delta 0.54 DNTT Terminal deoxynucleotidyltransferase 0.24 ELA2 Elastase 2, neutrophil 0.15 GATA3 GATA binding protein 3 0.81 MPO Myeloperoxidase 0.34 VPREB1 Pre-B lymphocyte gene 1 0.33

Fatty acid and lipid metabolism ACSL1,4,5 Acyl-CoA synthetase long-chain family members 1,4,5 1.29–1.50 FASN Fatty acid synthase 1.31 HADHA Hydroxyacyl-Coenzyme A dehydrogenase, alpha subunit 1.32

Receptors CD36 Thrombospondin receptor (CD36) 2.25 F2R Coagulation factor II (thrombin) receptor (PAR1) 1.45 FCGR2A Fc fragment of IgG receptor IIa (CD32) 1.40 GPR56 G protein-coupled receptor 56 1.61 LEPR Leptin receptor 4.12 P2RX5 Purinergic receptor P2X5 1.48 CCR2 Chemokine (C-C motif) receptor 2 0.68 CXCR4 Chemokine (C-X-C motif) receptor 4 0.65 INSR 0.76

Miscellaneous DLC1 Deleted in liver cancer 1 2.27 SRI Sorcin 1.40 TIMP1 Tissue inhibitor of metalloproteinase 1 1.82 SEPP1 Selenoprotein P, plasma, 1 0.44 The Q-value of each gene is o 0.1% (FDR 0.1%). Downregulated genes in CML are indicated by italics.

Next, we looked at BCR-ABL-activated mitogenic signaling Rac-1 and Rac-2, the downstream serine-threonine kinase pathways, such as Ras/MAPK, PI3 kinase and MYC pathways. Of mitogen-induced extracellular kinases 1 and 2 as well as the classical Ras/MAPK pathway Ras, the Ras-related proteins extracellular-regulated kinase (ERK) were significantly upregulated,

Leukemia Molecular signature of CML CD34 þ cells D-B Elena et al 498

INSR FGFR KIT BMPR CD44 VLA5 VLA4 1 Cytoplasmic membrane ITG A2B

ITG FLNA PXN PARV PAR1 surface PARV AE B A FLNB ACTG1 receptors FAK VCL CTNN CTNN CADH B1 BCR-ABL/adaptor protein complex A1 1 CD36 TLN1 Grb2 Tie-2 ShcDOK Crkl Cbl Crk cytoskeleton/

LEPR SOS Ras- adhesion/ Vav PI3K CADH GAP stem cell niche 2 CXCR 4 LYN Ras Rac1 Rac2 AKT CCR2

STAT5 STAT3,4,5 STAT1 NET1 Raf1 GCKR

BAD

HNRP SAPK p38 MEK1 MEK2 IRF8 K anti-apoptosis Bcl2 BclXL DUSP ERK JNK TGF  1 pathway

SMAD TGFB TGFB TFDP ELK1 Myc c-Jun JunD Atf-1 NFKB R1 2,4 1 Nucleus

differentiation Proliferation

Classical MAPK JNK MAPK p38 MAPK PI3 kinase pathway pathway pathway pathway

Figure 1 Differential expression of BCR-ABL substrates and genes involved in BCR-ABL-associated signaling pathways in primary CML CD34 þ hematopoietic stem and progenitor cells. Genes that were covered by our array are displayed by circular symbols. Abbreviations of genes (gene symbols) are explained in Table 1 and Supplementary Table 2. Significantly up- and downregulated genes are indicated by red and green, respectively, gray indicates no significantly differential expression. Arrows indicate activation of signaling proteins or transcriptional activation of genes, lines with crossbar indicate inhibition of signaling proteins.

whereas the ERK inhibitor dual-specific phosphatase 1 was in primary CML CD34 þ cells indicating that increased downregulated in primary BCR-ABL-positive cells. Of the signaling is not only based on phosphorylation but also on alternative c-Jun N-terminal kinase (JNK) MAPK pathway the transcriptional activation of signaling molecules. Our data germinal center kinase-related protein (GCKR; MAP4K5) and support the central role of the classical MAPK and the PI3 Rac-1 as well as Rac-2 were higher expressed, suggesting the kinase pathways for BCR-ABL-induced transformation of CML functional role of the MAPK pathways in BCR-ABL-positive CD34 þ cells. CD34 þ cells as previously shown for BCR-ABL-positive cell lines.26–28 The transcriptional target of MAPK pathways MYC was significantly upregulated, which emphasizes the central CML CD34 þ cells showed a higher expression level of role of transcriptional MYC activation in BCR-ABL-induced proliferation-associated genes and downregulation of transformation.29,30 The heterogenous nuclear ribonucleopro- the proapoptotic interferon regulatory factor 8 tein K (HNRPK), which positively regulates MYC expression Next, we looked at functional end points of BCR-ABL-signaling through ERK1/2 activation in blast crisis CML CD34 þ cells,31 such as proliferation, apoptosis and adhesion. In line with was also upregulated in chronic phase CML CD34 þ activation of mitogenic pathways, several genes encoding for cells, suggesting that HNRPK supports MAPK pathway-depen- cell cycle promoting proteins were upregulated in BCR-ABL- dent activation of MYC. Out of the genes involved in the PI3 positive CD34 þ cells such as the cell division cycle proteins kinase pathway, besides components of the PI3 kinase itself, its 16, 23, 27 and 37, the cyclins H, G1, the cyclin-dependent substrate AKT and the downstream molecules nuclear factor-kB kinases (CDK) 2, 7 and 8 as well as the CDK2-associated and BclXL were significantly upregulated in CML CD34 þ cells. protein. Moreover, the minichromosome maintenance protein Transcriptional activation of BclXL might be enhanced by 3, the origin recognition complex subunits 2, 4, 5, which are increased expression of STAT5 in leukemic progenitors.32,33 essential for initiation of DNA replication, and the DNA Our data suggest that survival and proliferation depends on polymerase epsilon 2 and 3 as well as the transcription factor the PI3 kinase pathway not only in BCR-ABL-transformed Dp-1 (TFDP1), which dimerizes with the E2F transcription factor murine cells and fibroblasts34,35 but also in primary CML and regulates the expression of various promoters involved in CD34 þ cells.36 cell cycle in normal CD34 þ cells,37,38 were also upregulated in In summary, several substrates of BCR-ABL as well as proteins CML progenitor cells. This finding is in line with the higher cell of BCR-ABL signaling pathways were significantly upregulated cycle activity of CML CD34 þ cells.39

Leukemia Molecular signature of CML CD34 þ cells D-B Elena et al 499 With respect to apoptosis-related proteins, we observed a respectively, were significantly downregulated in CML CD34 þ heterogenous picture. On the one hand, the proapoptotic key cells explaining previous findings of a reduced stromal-derived proteins caspases 3, 6 and 10 and the were factor-1 responsiveness of CML CD34 þ cells.51,52 significantly upregulated in BCR-ABL-positive CD34 þ cells. Thus, our expression data reflect the heterogenous results On the other hand, the antiapoptotic protein BclXL was also from previous studies with respect to the adhesion defect of CML upregulated in CML. Moreover, the interferon regulatory factor 8 cells. In summary, our results suggest that the release of (IRF8, interferon consensus sequence binding protein) was progenitor cells in CML into PB might be supported by a significantly downregulated in primary CML CD34 þ cells. downregulation of the chemokine receptors CXCR4 and CCR2 IRF8 could induce apoptosis in BCR-ABL-positive myeloid cells as well as the adhesion molecules L-selectin and N-cadherin. in mice40 and IRF8 knock-out mice developed a myeloproli- ferative syndrome.41 Recently, it has been shown in BCR-ABL- transformed murine cells that IRF8 antagonizes BCR-ABL by Upregulation of genes of the TGFb signaling pathway downregulation of bcl-2 and could override imatinib resis- Activation of BCR-ABL-induced mitogenic signaling can be tance.42 Thus, our data derived from CD34 þ cells support the enhanced by upregulation of growth factors and their receptors. hypothesis that BCR-ABL-induced inhibition of apoptosis in Indeed, upregulation of macrophage colony-stimulating factor, primary human leukemic progenitor cells is mediated by IL1b, IL7 and transforming growth factor (TGF)b1 as well as of downregulation of IRF8. the granulocyte–macrophage colony-stimulating factor receptor was observed. In turn, IL7 receptor and FLT3 were significantly downregulated in leukemic progenitor cells. Looking in detail at CML CD34 þ cells had an increased expression of focal the TGFb signaling pathway TGFb1 itself as well as SMAD2, adhesion proteins and a reduced expression of L-selectin SMAD4 and the TFDP1 were significantly upregulated in CML and N-cadherin. CD34 þ cells. TGFb1 generally suppresses stem cell prolifera- CML progenitor cells have a decreased capability to adhere to tion and has inhibitory effects on cytokine-induced cell growth BM stroma4,5 and are released from BM to PB to a larger extent in chronic phase CML.53 On the other hand, autocrine TGFb in comparison to normal progenitor cells. The reasons for this stimulates early erythropoiesis and reactivates fetal hemoglobin phenomenon are not yet understood. Therefore, we especially (Hb) synthesis in adult erythroid progenitor cells.54,55 As looked on differential expression of genes involved in progenitor erythropoiesis-associated genes and fetal globin genes were cell adhesion and migration. The BCR-ABL adapter protein Crkl, also upregulated in CML progenitor cells (see below), we which facilitates complex formation with focal adhesion hypothesize that erythropoiesis and fetal Hb synthesis is proteins and is involved in regulation of cellular motility43 activated in chronic phase CML by an autocrine stimulation of was upregulated in CML CD34 þ cells. Among the Crkl- the TGFb pathway. Interestingly, the neuroepithelial cell associated adhesion proteins vinculin, talin, parvin and filamin transforming gene 1 (NET1), which is induced upon TGFb were also significantly upregulated in CML (Figure 1). Previous stimulation56 is also upregulated in CML CD34 þ cells. NET1 studies reported alterations in integrin-mediated cell-adhesion plays a central role in Rho GTPase (RAC1)-mediated reorgani- of CML cells.5,44 Looking at changes in integrin expression, we zation of the cytoskeleton57 and activation of the alternative JNK found only the integrins aE and a2b to be upregulated. MAPK pathway.58 Therefore, TGFb-induced NET1 upregulation Therefore, there is no evidence for transcriptional activation of might enhance the MAPK-mediated mitogenic activation of most of the integrins in CML CD34 þ cells. However, BCR-ABL CML CD34 þ cells. could alter integrin-mediated adhesion and signaling by modulation of the functional state rather than by the expression level of the integrin receptors. A finding along this line was the CML CD34 þ cells showed an expression pattern significantly higher expression of the SRC kinase LYN in CML characteristic for immature stem cells as well as CD34 þ cells. LYN positively regulates CD34 þ cell movement megakaryocyte-erythrocyte progenitor cells and lowers adhesion of CD34 þ cells to stromal cells by Further, we looked at genes involved in early hematopoiesis and inhibiting the ICAM-1-binding activity of b2 integrins.45 differentiation. We found several genes associated with early Increased expression of LYN might therefore be involved in HSCs and their self-renewal to be significantly upregulated in the release of CML progenitor cells into the PB. Additionally, CML CD34 þ cells indicating a molecular phenotype of a more overexpression of LYN was associated with imatinib resis- primitive hematopoietic progenitor cell. These genes include tance.46 Thus, the upregulation of LYN in CML CD34 þ stem GATA2, homeo box genes HoxA9 and HoxA10, MEIS1, zinc and progenitor cells provides the rational for using the finger protein Ikaros, Tal-1, lim-finger protein Lmo-2, member of combined ABL/SRC kinase inhibitor dasatinib even in therapy the Polycomb group Bmi-1 and aldehyd dehygrogenase naı¨ve patients with CML. Interestingly, several proteins involved ALDH1A1, which is expressed in long-term reconstituting HSCs in regulation of the actin cytoskeleton were deregulated in CML and considered to be involved in chemoresistance.30,59,60 Thus, CD34 þ cells in comparison with normal cells. The BCR-ABL our gene expression data suggest a greater self-renewal capacity associated cytoskeletal protein, actin-g1, a component of f-actin, of leukemic CD34 þ cells. Overexpression of HoxA9, MEIS1 that regulates cell adhesion47,48 had a significantly lower and PBX3 might be relevant for the malignant transformation of expression level, whereas the regulators of the actin cytoskele- CML CD34 þ cells, as it has recently been shown that Hox- ton, Rho GTPases Rac1 and Rac2, were significantly upregu- MEIS-PBX complexes drive leukemogenesis in AML.61 lated in CML CD34 þ cells.49 Additional adhesion molecules, Looking at genes involved in differentiation, we found that ICAM-4, cadherin 1 (E-cadherin) and the cadherin-associated GATA1, associated with early myeloid differentiation, was protein catenin-a1 were upregulated and L-selectin and cadher- significantly upregulated and myeloperoxidase, neutrophil in 2 (N-cadherin), a molecule involved in the interaction of the elastase 2 (ELA2) and C/EBPd, genes expressed during later stem cell and its niche,50 were downregulated in CML CD34 þ stages of myeloid differentiation, were downregulated in CML cells. Finally, the chemokine receptors CXCR4 and CCR2, CD34 þ cells. Two of the central genes promoting lymphoid which are involved in stem cell and monocyte migration, differentiation, GATA3 and IL-7 receptor a, as well as the B-cell

Leukemia Molecular signature of CML CD34 þ cells D-B Elena et al 500 differentiation-associated genes terminal deoxynucleotidyltrans- We found several genes involved in fatty acid and lipid ferase, pre-B lymphocyte gene 1 (VPREB1), B-cell linker (BLNK) metabolism to be upregulated in CML CD34 þ cells. For and BCL-6 were downregulated in CML CD34 þ cells. On the example, the isoenzymes of the long-chain fatty-acid-coenzyme other hand, erythropoiesis-associated genes such as erythro- A ligases ACSL 1, 4 and 5, were higher expressed in leukemic poietin receptor, Kruppel-like factor 1 and transferrin receptor CD34 þ cells in comparison to normal cells. As ACSL play a key were higher expressed in CML CD34 þ cells. An interesting role in lipid biosynthesis and are expressed in erythrocyte finding was an upregulation of fetal Hb components such as precursors71 our expression data might reflect a higher propor- Hbg1, Hbg2 and Hbd as well as Hbb, whereas the genes for tion of erythroid progenitors in CML CD34 þ cells. Furthermore, adult Hb chains Hba1 and Hba2 were not altered indicating an fatty acid synthase (FASN), which catalyzes the synthesis of induction of fetal Hb synthesis in CML progenitor cells. The re- long-chain saturated fatty acids, and hydroxyacyl-coenzyme A activation of fetal Hb synthesis in CML progenitors could be dehydrogenasea, which promotes oxidation of fatty acids, were induced by an autocrine activation of the TGFb signaling also upregulated in CML. FASN was overexpressed in aggressive pathway (see above).54,55 In summary, the gene expression breast cancer and inhibition of fatty acid synthesis resulted in signature of CML CD34 þ cells indicates a molecular pheno- cell cycle arrest and apoptosis of tumor cell lines.72 Thus, a type of early HSCs with preponderance for megakaryocyte- pharmacological inhibitor of FASN such as cerulenin could be erythroid progenitors rather than for granulocyte–macrophage useful for inhibition of leukemic cell growth in CML. progenitors (GMP). Corroboration by quantitative real-time RT-PCR The receptor for leptin was upregulated in For corroboration of the microarray data, we assessed the mRNA CML CD34 þ cells expression levels of 8 genes. We did not find any discrepancy of Several surface receptors, which have not been mentioned in array and RT-PCR data and confirmed significantly (Po0.05) the context of the previously addressed functional groups were differential expression of TAL1, LEPR, GATA1, TIMP1, CRHBP, also deregulated in CML CD34 þ cells. As an example, the ELA2, FACL5, SEPP1 (Figure 2). LEPR, a member of the cytokine receptor superfamily, which signals via activation of JAK2, STAT1, 3 and 5 proteins as well as CML CD34 þ cells have a greater proportion of MEP but MAPK pathways and plays an important role for several less HSC and GMP in comparison with normal CD34 þ physiological and pathophysiological processes such as body cells mass control, metabolic pathways, reproduction, angiogenesis, In order to address the question whether the differential bone remodelling and immunity62 showed the greatest level of expression of stem cell- and differentiation-associated genes is differential expression in CML CD34 þ cells. In hematopoiesis due to a different composition of normal and CML CD34 þ leptin induces proliferation of hematopoietic cells and stimu- cells, we performed a detailed analysis of stem and progenitor lates the differentiation of lineage-restricted precursors of the cell subsets. The proportion of early HSCs was significantly erythrocytic and myelopoietic lineages.63 LEPR has also smaller in CML CD34 þ cells compared to their normal proliferative and antiapoptotic activities in AML64 and had a counterparts (3.6%; s.d.: 2.1 vs 12.7%; s.d.: 2.9%; P ¼ 0.011) high expression level in CML blast crisis65 suggesting a (Figure 3). Within the lin-, CD38 þ subfraction of CML CD34 þ pathophysiological role in leukemia. As LEPR is upregulated in cells a greater proportion of megakaryocyte-erythrocyte pro- primary CML CD34 þ cells this receptor might activate genitor cells (MEP) was observed in comparison with normal proliferation and differentiation of leukemic progenitors and CD34 þ cells (47.2%; s.d.: 10.2 vs 17.1%; s.d.: 4.8%; P ¼ 0.01), might therefore be a suitable target for novel therapeutic approaches.66 Further differentially expressed surface receptors are shown in Table 1. 6 2 Upregulation of genes involved in fatty-acid metabolism 4 and in the ubiquitin-proteasome system Other interesting genes, which were upregulated in CML 2 ELA2 SEPP1 CD34 þ cells, were sorcin that is related to drug resistance CRHBP and poor prognosis in AML, and the candidate tumor suppressor 0 gene DLC1 (deleted in liver cancer 1), which activates Rho

67 -2 TAL1 LEPR TIMP1

GTPases and modulates the organization of the cytoskeleton. GATA1 FACL5 Moreover, the TIMP1 was also upregulated in CML CD34 þ cells. TIMP1 mediates JAK2-, AKT- and PI3 kinase-dependent -4 survival of erythroid cells and is upregulated in leukemic blasts.68,69 Hence, TIMP1 overexpression might enhance -6 Differential gene expression log activation of PI3 kinase-AKT pathway in CML progenitor cells. -8 The SEPP1, which is involved in response to oxidative stress, was downregulated in CML CD34 þ cells. In addition, several Figure 2 Corroboration of microarray expression data by RT-PCR. genes involved in the ubiquitin-proteasome system, such as the Fold changes determined using RT-PCR are shown in white columns, ubiquitin-conjugating enzymes E2M, E2G1, E2G2, E2G3, E2E1, fold changes assessed by the SAM software from microarray analysis E2N, E2D2, E2D3, the SUMO-1 activating enzyme (UBA2), the are indicated in black columns. Fold changes 40 indicate higher expression in CML CD34 þ cells, fold changes 0 indicate higher F-box protein 7, UBADC1, and the proteasome subunits a-1, -2, o expression in normal CD34 þ cells. Y axis is scaled according to log -3, -4, -7 and subunits b-1, -2, -5 were significantly upregulated. base 2. Expression of all genes assessed by RT-PCR was significantly This supports a previous finding that BCR-ABL induces a different (Po0.05). Results from RT-PCR were based on five samples proteasome-mediated degradation of inhibitory proteins.70 from CML and five samples from healthy volunteers.

Leukemia Molecular signature of CML CD34 þ cells D-B Elena et al 501 103 103

HSC Progenitors HSC Progenitors 102 102 12.8% 57.2% 1.3% 74.7%

101 101 CD34-PerCP CD34-PerCP 100 100

1 2 3 100 10 10 10 100 101 102 103 CD38-APC CD38-APC 103 103 CMP 29.6% 102 102 3.8% CMP 0.2% 32.4% -PE -PE 1

 1  10 GMP 10 GMP

42.1% IL-3 26.8% IL-3

100 100 MEP MEP 22.5% 39.7%

100 101 102 103 100 101 102 103 CD45RA-ECD CD45RA-ECD

Figure 3 Different proportions of hematopoietic stem and progenitor cells in BM of either healthy volunteers (left) or patients with CML (right). Immunomagnetic selection of CD34 þ cells followed by multi-color flow cytometry was utilized to analyze hematopoietic stem and progenitor cell subsets. After gating on viable cells and lineage-depletion subfractions of HSC (Lin-, CD34 þ , CD38-), CMP (Lin-, CD34 þ , CD38 þ , IL- 3Ra þ , CD45RA-), GMP (Lin-, CD34 þ , CD38 þ , IL-3Ra þ , CD45RA þ ), MEP (Lin-, CD34 þ , CD38 þ , IL-3Ra-, CD45RA-) and IL-3 receptor ahigh progenitors (Lin-, CD34 þ , CD38 þ , IL-3Rahigh CD45RA þ ) were determined. Progenitor subset analysis was performed following gating on lin-, CD38 þ cells. Percentages of subsets referring to all cells in the respective dot blot are indicated. whereas the proportions of GMP cells and the IL-3 receptor a- extremely high neutrophil counts and almost normal Hb values high progenitor population was significantly smaller in CML in patients with CML is not clear. One might speculate that there (GMP: 24.7%; s.d.: 5.8 vs 46.4%; s.d.: 3.7%; p ¼ 0.005; is either a differentiation block of late erythropoietic progenitors, IL3Rahigh 0.1%; s.d. 0.1 vs 6.6%; s.d. 2.6%; P ¼ 0.01). a hypothesis that was already published in the late 1970s,73 or Proportions of common myeloid progenitor cells (CMP) were an increase of early erythropoiesis and a reduced proliferation of not different (27.8%; s.d.: 13.7 vs 27.1%; s.d.: 5.0%). This is in late erythroid precursors. The latter hypothesis could be line with our gene expression data, which showed a higher explained by the upregulation of components of the TGFb expression level of erythropoiesis-associated genes and which pathway in CML CD34 þ cells (see above), as it has been exhibited a lower expression level of late myelopoiesis- previously shown that TGFb1 is a paradoxical inhibitor of associated genes. The results also show that the expansion of erythropoiesis that acts by blocking proliferation and accelerat- malignant precursor cells in chronic phase CML does not occur ing early differentiation.74 On the other hand, expansion of cells on HSC level. Additionally, the higher expression of stem cell- of the myeloid lineage in chronic phase CML might occur on a associated genes such as GATA2 or the Hox genes in CML late neutrophilic precursor level, whereas early progenitor cells CD34 þ cells in combination with a smaller proportion of HSC fail to expand due to a higher probability of differentiation as suggests that those genes are transcriptionally activated in CML already suggested in a previous paper.75 HSC and might result in an increased self-renewal capacity of those cells. Our results suggest that the up- and downregulation of Conclusions differentiation and stem cell-associated genes in CD34 þ cells are due to both, a different subset composition as well as an In this study, we provide for the first time a genome-wide gene upregulation of these genes in the respective progenitor cell expression signature of highly enriched CD34 þ hematopoietic subset. This shows that although highly enriched CD34 þ cells stem and progenitor cells from patients with CML in chronic have been used for gene expression analysis, we are still looking phase. Our data show that several of the BCR-ABL-induced at data from a rather heterogenous cell population. transcriptional effects described in cell lines and BCR-ABL- The reason for the increased proportion of erythroid transduced murine cells can also be found in primary CML progenitors (MEP) and upregulation of erythropoiesis-associated progenitor cells. Most components of the classical MAPK genes as well as the reduced proportion of GMP and IL3Rahigh pathway and the PI3 kinase/AKT pathway and some genes of progenitors in CML CD34 þ cells in combination with the alternative JNK and p38 MAPK pathways were upregulated

Leukemia Molecular signature of CML CD34 þ cells D-B Elena et al 502 in primary CML CD34 þ cells. This result also shows, that 10 Ohmine K, Ota J, Ueda M, Ueno S, Yoshida K, Yamashita Y et al. activation of those pathways not only occurs by phosphorylation Characterization of stage progression in chronic myeloid leukemia of proteins but also on a transcriptional level. With respect to by DNA microarray with purified hematopoietic stem cells. genes involved in adhesion and migration, we found a Oncogene 2001; 20: 8249–8257. 11 Frank O, Brors B, Fabarius A, Li L, Haak M, Merk S et al. Gene heterogenous picture that reflects the results from previous 3 expression signature of primary imatinib-resistant chronic myeloid studies. Moreover, several novel transcriptional changes in leukemia patients. Leukemia 2006; 20: 1400–1407. comparison with normal CD34 þ cells, which were not 12 McLean LA, Gathmann I, Capdeville R, Polymeropoulos MH, described so far in CML and which might be of therapeutic Dressman M. Pharmacogenomic analysis of cytogenetic response relevance, were identified. These include an upregulation of in chronic myeloid leukemia patients treated with imatinib. Clin components of the TGFb signaling pathway, a higher expression Cancer Res 2004; 10: 155–165. 13 Neumann F, Teutsch N, Kliszewski S, Bork S, Steidl U, Brors B level of fetal Hb genes, an upregulation of genes involved in et al. Gene expression profiling of Philadelphia (Ph)- fatty acid and lipid metabolism or interesting candidate genes, negative CD34+ hematopoietic stem and progenitor cells of such as the LEPR, thrombin receptor (PAR1) or NET1. patients with Ph-positive CML in major molecular remission Additionally, our expression data suggested an altered composi- during therapy with imatinib. Leukemia 2005; 19: 458–460. tion of the CD34 þ cell population in CML with an increase of 14 Villuendas R, Steegmann JL, Pollan M, Tracey L, Granda A, erythroid progenitors and a decrease of late myeloid progenitor Fernandez-Ruiz E et al. Identification of genes involved in imatinib cells, which we could confirm by subset analysis of CML resistance in CML: a gene-expression profiling approach. Leuke- mia 2006; 20: 1047–1054. CD34 þ cells. Therefore, one has to keep in mind that the 15 Yong AS, Szydlo RM, Goldman JM, Apperley JF, Melo JV. expression profile reflects at least partially the different Molecular profiling of CD34+ cells identifies low expression of proportions of cell subsets in the CD34 þ cell population. CD7, along with high expression of proteinase 3 or elastase, as However, our data can form the basis for future studies on predictors of longer survival in patients with CML. Blood 2006; highly enriched CD34 þ lin-CD38- HSC and might help to 107: 205–212. identify novel stem or progenitor cell-associated targets for 16 Radich JP, Dai H, Mao M, Oehler V, Schelter J, Druker B et al. Gene expression changes associated with progression and supplementation of therapy with imatinib. response in chronic myeloid leukemia. Proc Natl Acad Sci USA 2006; 103: 2794–2799. 17 Zheng C, Li L, Haak M, Brors B, Frank O, Giehl M et al. Gene Acknowledgements expression profiling of CD34+ cells identifies a molecular signature of chronic myeloid leukemia blast crisis. Leukemia Following financial supports are greatly appreciated: ED-B: 2006; 20: 1028–1034. Deutscher Akademischer Austauschdienst (DAAD), MR: Scholar- 18 Steidl U, Kronenwett R, Rohr UP, Fenk R, Kliszewski S, Maercker ship of Duesseldorf University, BB: National Genome Research C et al. Gene expression profiling identifies significant differences Network (grant 01 GR 0450), SB: Leuka¨mie Liga eV Du¨sseldorf, between the molecular phenotypes of bone marrow-derived and circulating human CD34+ hematopoietic stem cells. Blood 2002; Forschungskommission der Medizinischen Fakulta¨tDu¨sseldorf, 99: 2037–2044. AvH: Vienna Science and Technology Fund (WWTF). 19 Huber W, von Heydebreck A, Sultmann H, Poustka A, Vingron M. Variance stabilization applied to microarray data calibration and to the quantification of differential expression. Bioinformatics 2002; 18 Suppl 1: S96–S104. References 20 Tusher VG, Tibshirani R, Chu G. Significance analysis of microarrays applied to the ionizing radiation response. Proc Natl 1 Faderl S, Talpaz M, Estrov Z, O’Brien S, Kurzrock R, Kantarjian Acad Sci USA 2001; 98: 5116–5121. HM. The biology of chronic myeloid leukemia. N Engl J Med 1999; 21 Kronenwett R, Graf T, Nedbal W, Weber M, Steidl U, 341: 164–172. Rohr UP et al. Inhibition of angiogenesis in vitro by alphav 2 Daley GQ, Van Etten RA, Baltimore D. Induction of chronic integrin-directed antisense oligonucleotides. Cancer Gene Ther myelogenous leukemia in mice by the P210bcr/ gene of the 2002; 9: 587–596. Philadelphia chromosome. Science 1990; 247: 824–830. 22 Manz MG, Miyamoto T, Akashi K, Weissman IL. Prospective 3 Deininger MW, Goldman JM, Melo JV. The molecular biology of isolation of human clonogenic common myeloid progenitors. Proc chronic myeloid leukemia. Blood 2000; 96: 3343–3356. Natl Acad Sci USA 2002; 99: 11872–11877. 4 Gordon MY, Dowding CR, Riley GP, Goldman JM, Greaves MF. 23 Brazma A, Hingamp P, Quackenbush J, Sherlock G, Spellman P, Altered adhesive interactions with marrow stroma of haemato- Stoeckert C et al. Minimum information about a microarray poietic progenitor cells in chronic myeloid leukaemia. Nature experiment (MIAME)-toward standards for microarray data. Nat 1987; 328: 342–344. Genet 2001; 29: 365–371. 5 Salesse S, Verfaillie CM. Mechanisms underlying abnormal 24 Staal FJ, Cario G, Cazzaniga G, Haferlach T, Heuser M, Hofmann trafficking and expansion of malignant progenitors in CML: BCR/ WK et al. Consensus guidelines for microarray gene expression ABL-induced defects in integrin function in CML. Oncogene 2002; analyses in leukemia from three European leukemia networks. 21: 8605–8611. Leukemia 2006; 20: 1385–1392. 6 Marley SB, Gordon MY. Chronic myeloid leukaemia: stem cell 25 Ng YY, van Kessel B, Lokhorst HM, Baert MR, van den Burg CM, derived but progenitor cell driven. Clin Sci (London) 2005; 109: Bloem AC et al. Gene-expression profiling of CD34+ cells 13–25. from various hematopoietic stem-cell sources reveals functional 7 Bhatia R, Holtz M, Niu N, Gray R, Snyder DS, Sawyers CL et al. differences in stem-cell activity. J Leukocyte Biol 2004; 75: Persistence of malignant hematopoietic progenitors in chronic 314–323. myelogenous leukemia patients in complete cytogenetic remission 26 Cortez D, Reuther G, Pendergast AM. The Bcr-Abl tyrosine kinase following imatinib mesylate treatment. Blood 2003; 101: activates mitogenic signaling pathways and stimulates G1-to-S 4701–4707. phase transition in hematopoietic cells. Oncogene 1997; 15: 8 Kronenwett R, Butterweck U, Steidl U, Kliszewski S, Neumann F, 2333–2342. Bork S et al. Distinct molecular phenotype of malignant CD34(+) 27 Raitano AB, Halpern JR, Hambuch TM, Sawyers CL. The Bcr-Abl hematopoietic stem and progenitor cells in chronic myelogenous leukemia oncogene activates Jun kinase and requires Jun for leukemia. Oncogene 2005; 24: 5313–5324. transformation. Proc Natl Acad Sci USA 1995; 92: 11746–11750. 9 Nowicki MO, Pawlowski P, Fischer T, Hess G, Pawlowski T, 28 Shi CS, Tuscano JM, Witte ON, Kehrl JH. GCKR links the Bcr-Abl Skorski T. Chronic myelogenous leukemia molecular signature. oncogene and Ras to the stress-activated protein kinase pathway. Oncogene 2003; 22: 3952–3963. Blood 1999; 93: 1338–1345.

Leukemia Molecular signature of CML CD34 þ cells D-B Elena et al 503 29 Sawyers CL, Callahan W, Witte ON. Dominant negative MYC 49 Tapon N, Hall A. Rho, Rac and Cdc42 GTPases regulate the blocks transformation by ABL oncogenes. Cell 1992; 70: 901–910. organization of the actin cytoskeleton. Curr Opin Cell Biol 1997; 30 Zhu J, Emerson SG. Hematopoietic cytokines, transcription factors 9: 86–92. and lineage commitment. Oncogene 2002; 21: 3295–3313. 50 Adams GB, Scadden DT. The hematopoietic stem cell in its place. 31 Notari M, Neviani P, Santhanam R, Blaser BW, Chang JS, Galietta Nat Immunol 2006; 7: 333–337. A et al. A MAPK/HNRPK pathway controls BCR/ABL oncogenic 51 Durig J, Rosenthal C, Elmaagacli A, Heyworth C, Halfmeyer K, potential by regulating MYC mRNA translation. Blood 2006; 107: Kasper C et al. Biological effects of stroma-derived factor-1 alpha 2507–2516. on normal and CML CD34+ haemopoietic cells. Leukemia 2000; 32 Horita M, Andreu EJ, Benito A, Arbona C, Sanz C, Benet I et al. 14: 1652–1660. Blockade of the Bcr-Abl kinase activity induces apoptosis of 52 Peled A, Hardan I, Trakhtenbrot L, Gur E, Magid M, Darash- chronic myelogenous leukemia cells by suppressing signal Yahana M et al. Immature leukemic CD34+CXCR4+ cells from transducer and activator of transcription 5-dependent expression CML patients have lower integrin-dependent migration and of Bcl-xL. J Exp Med 2000; 191: 977–984. adhesion in response to the chemokine SDF-1. Stem Cells 2002; 33 Sillaber C, Gesbert F, Frank DA, Sattler M, Griffin JD. STAT5 20: 259–266. activation contributes to growth and viability in Bcr/Abl-trans- 53 Murohashi I, Endho K, Nishida S, Yoshida S, Jinnai I, Bessho M formed cells. Blood 2000; 95: 2118–2125. et al. Differential effects of TGF-beta 1 on normal and leukemic 34 Skorski T, Bellacosa A, Nieborowska-Skorska M, Majewski M, human hematopoietic cell proliferation. Exp Hematol 1995; 23: Martinez R, Choi JK et al. Transformation of hematopoietic cells by 970–977. BCR/ABL requires activation of a PI-3k/Akt-dependent pathway. 54 Bhanu NV, Trice TA, Lee YT, Gantt NM, Oneal P, Schwartz JD EMBO J 1997; 16: 6151–6159. et al. A sustained and pancellular reversal of gamma-globin gene 35 Skorski T, Kanakaraj P, Nieborowska-Skorska M, Ratajczak MZ, silencing in adult human erythroid precursor cells. Blood 2005; WenSC,ZonGet al. Phosphatidylinositol-3 kinase activity is 105: 387–393. regulated by BCR/ABL and is required for the growth of Philadelphia 55 Bohmer RM, Campbell TA, Bianchi DW. Selectively increased chromosome-positive cells. Blood 1995; 86: 726–736. growth of fetal hemoglobin-expressing adult erythroid progenitors 36 Andreu EJ, Lledo E, Poch E, Ivorra C, Albero MP, Martinez-Climent after brief treatment of early progenitors with transforming growth JA et al. BCR-ABL induces the expression of Skp2 through the factor beta. Blood 2000; 95: 2967–2974. PI3 K pathway to promote p27Kip1 degradation and proliferation 56 Shen X, Li J, Hu PP, Waddell D, Zhang J, Wang XF. The activity of of chronic myelogenous leukemia cells. Cancer Res 2005; 65: guanine exchange factor NET1 is essential for transforming growth 3264–3272. factor-beta-mediated stress fiber formation. J Biol Chem 2001; 276: 37 Williams CD, Linch DC, Watts MJ, Thomas NS. Characterization 15362–15368. of cell cycle status and E2F complexes in mobilized CD34+ 57 Qin H, Carr HS, Wu X, Muallem D, Tran NH, Frost JA. cells before and after cytokine stimulation. Blood 1997; 90: Characterization of the biochemical and transforming properties 194–203. of the neuroepithelial transforming protein 1. J Biol Chem 2005; 38 Wu CL, Classon M, Dyson N, Harlow E. Expression of dominant- 280: 7603–7613. negative mutant DP-1 blocks cell cycle progression in G1. Mol 58 Alberts AS, Treisman R. Activation of RhoA and SAPK/JNK Cell Biol 1996; 16: 3698–3706. signalling pathways by the RhoA-specific exchange factor mNET1. 39 Primo D, Flores J, Quijano S, Sanchez ML, Sarasquete ME, EMBO J 1998; 17: 4075–4085. Pino-Montes J et al. Impact of BCR/ABL gene expression on the 59 Pineault N, Helgason CD, Lawrence HJ, Humphries RK. Differ- proliferative rate of different subpopulations of haematopoietic ential expression of Hox, Meis1, and Pbx1 genes in primitive cells cells in chronic myeloid leukaemia. Br J Haematol 2006; 135: throughout murine hematopoietic ontogeny. Exp Hematol 2002; 43–51. 30: 49–57. 40 Hao SX, Ren R. Expression of interferon consensus sequence 60 Storms RW, Trujillo AP, Springer JB, Shah L, Colvin OM, Ludeman binding protein (ICSBP) is downregulated in Bcr-Abl-induced SM et al. Isolation of primitive human hematopoietic progenitors murine chronic myelogenous leukemia-like disease, and forced on the basis of aldehyde dehydrogenase activity. Proc Natl Acad coexpression of ICSBP inhibits Bcr-Abl-induced myeloproliferative Sci USA 1999; 96: 9118–9123. disorder. Mol Cell Biol 2000; 20: 1149–1161. 61 Wang GG, Pasillas MP, Kamps MP. Persistent transactivation by 41 Holtschke T, Lohler J, Kanno Y, Fehr T, Giese N, Rosenbauer F meis1 replaces hox function in myeloid leukemogenesis models: et al. Immunodeficiency and chronic myelogenous leukemia-like evidence for co-occupancy of meis1-pbx and hox-pbx complexes syndrome in mice with a targeted mutation of the ICSBP gene. Cell on promoters of leukemia-associated genes. Mol Cell Biol 2006; 1996; 87: 307–317. 26: 3902–3916. 42 Burchert A, Cai D, Hofbauer LC, Samuelsson MK, Slater EP, 62 Fruhbeck G. Intracellular signalling pathways activated by leptin. Duyster J et al. Interferon consensus sequence binding protein Biochem J 2006; 393: 7–20. (ICSBP; IRF-8) antagonizes BCR/ABL and down-regulates bcl-2. 63 Gainsford T, Willson TA, Metcalf D, Handman E, McFarlane C, Ng Blood 2004; 103: 3480–3489. A et al. Leptin can induce proliferation, differentiation, and 43 Uemura N, Griffin JD. The adapter protein Crkl links Cbl to C3G functional activation of hemopoietic cells. Proc Natl Acad Sci after integrin ligation and enhances cell migration. J Biol Chem USA 1996; 93: 14564–14568. 1999; 274: 37525–37532. 64 Konopleva M, Mikhail A, Estrov Z, Zhao S, Harris D, 44 Bazzoni G, Carlesso N, Griffin JD, Hemler ME. Bcr/Abl expression Sanchez-Williams G et al. Expression and function of leptin stimulates integrin function in hematopoietic cell lines. J Clin receptor isoforms in myeloid leukemia and myelodysplastic Invest 1996; 98: 521–528. syndromes: proliferative and anti-apoptotic activities. Blood 45 Nakata Y, Tomkowicz B, Gewirtz AM, Ptasznik A. Integrin 1999; 93: 1668–1676. inhibition through Lyn-dependent cross talk from CXCR4 chemo- 65 Nakao T, Hino M, Yamane T, Nishizawa Y, Morii H, Tatsumi N. kine receptors in normal human CD34+ marrow cells. Blood Expression of the leptin receptor in human leukaemic blast cells. 2006; 107: 4234–4239. Br J Haematol 1998; 102: 740–745. 46 Donato NJ, Wu JY, Stapley J, Gallick G, Lin H, Arlinghaus R et al. 66 Paraskevas KI, Liapis CD, Mikhailidis DP. Leptin: a promising BCR-ABL independence and LYN kinase overexpression in therapeutic target with pleiotropic action besides body weight chronic myelogenous leukemia cells selected for resistance to regulation. Curr Drug Targets 2006; 7: 761–771. STI571. Blood 2003; 101: 690–698. 67 Durkin ME, Avner MR, Huh CG, Yuan BZ, Thorgeirsson SS, 47 McWhirter JR, Galasso DL, Wang JY. A coiled-coil oligomerization Popescu NC. DLC-1, a Rho GTPase-activating protein with tumor domain of Bcr is essential for the transforming function of Bcr-Abl suppressor function, is essential for embryonic development. FEBS oncoproteins. Mol Cell Biol 1993; 13: 7587–7595. Lett 2005; 579: 1191–1196. 48 Wertheim JA, Perera SA, Hammer DA, Ren R, Boettiger D, 68 Janowska-Wieczorek A, Marquez LA, Matsuzaki A, Hashmi HR, Pear WS. Localization of BCR-ABL to F-actin regulates cell Larratt LM, Boshkov LM et al. Expression of matrix metalloprotei- adhesion but does not attenuate CML development. Blood 2003; nases (MMP-2 and -9) and tissue inhibitors of metalloproteinases 102: 2220–2228. (TIMP-1 and -2) in acute myelogenous leukaemia blasts: compar-

Leukemia Molecular signature of CML CD34 þ cells D-B Elena et al 504 ison with normal bone marrow cells. Br J Haematol 1999; 105: 72 Pizer ES, Chrest FJ, DiGiuseppe JA, Han WF. Pharmacological 402–411. inhibitors of mammalian fatty acid synthase suppress DNA 69 Lambert E, Boudot C, Kadri Z, Soula-Rothhut M, Sowa ML, replication and induce apoptosis in tumor cell lines. Cancer Res Mayeux P et al. Tissue inhibitor of metalloproteinases-1 signalling 1998; 58: 4611–4615. pathway leading to erythroid cell survival. Biochem J 2003; 372: 73 Eaves AC, Eaves CJ. Abnormalities in the erythroid progenitor 767–774. compartments in patients with chronic myelogenous leukemia 70 Dai Z, Quackenbush RC, Courtney KD, Grove M, Cortez D, (CML). Exp Hematol 1979; 7 Suppl 5: 65–75. Reuther GW et al. Oncogenic Abl and Src tyrosine kinases elicit 74 Zermati Y, Fichelson S, Valensi F, Freyssinier JM, Rouyer-Fessard the ubiquitin-dependent degradation of target proteins through a P, Cramer E et al. Transforming growth factor inhibits erythropoi- Ras-independent pathway. Genes Dev 1998; 12: 1415–1424. esis by blocking proliferation and accelerating differentiation of 71 Malhotra KT, Malhotra K, Lubin BH, Kuypers FA. Identification erythroid progenitors. Exp Hematol 2000; 28: 885–894. and molecular characterization of acyl-CoA synthetase in 75 Eaves AC, Barnett MJ, Ponchio L, Cashman JD, Petzer AL, Eaves CJ. human erythrocytes and erythroid precursors. Biochem J 1999; Differences between normal and CML stem cells: potential targets 344 (Part 1): 135–143. for clinical exploitation. Stem Cells 1998; 16 Suppl 1: 77–83.

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

Leukemia