Distinct Molecular Phenotype of Malignant Cd34þ Hematopoietic

Distinct molecular phenotype of malignant CD34 þ hematopoietic stem and progenitor cells in chronic myelogenous leukemia

Ralf Kronenwett*,1, Ulf Butterweck1, Ulrich Steidl1,2, Slawomir Kliszewski1, Frank Neumann1, Simone Bork1, Elena Diaz Blanco1, Nicole Roes1, Thorsten Gra¨ f1, Benedikt Brors3, Roland Eils3, Christian Maercker4, Guido Kobbe1, Norbert Gattermann1 and Rainer Haas1

1Department of Hematology, Oncology and Clinical Immunology, Heinrich Heine University Duesseldorf, Moorenstr. 5, 40225 Duesseldorf, Germany; 2Harvard Institutes of Medicine, Hematology/Oncology Division, Boston, MA, USA; 3Theoretical Bioinformatics, German Cancer Research Center, 69120 Heidelberg, Germany; 4German Resource Center for Genome Research, 69120 Heidelberg, Germany

Chronic myelogenous leukemia (CML) is a malignant Keywords: CML; CD34 þ cells; gene expression; G disorder of the hematopoietic stem cell characterized by protein-coupled receptors the BCR–ABL oncogene. We examined gene expression profiles of highly enriched CD34 þ hematopoietic stem and progenitor cells from patients with CML in chronic phase using cDNA arrays covering 1.185 genes. Compar- ing CML CD34 cells with normal CD34 cells, we þ þ Introduction found 158 genes which were significantly differentially expressed. Gene expression patterns reflected BCR–ABL- Hematopoietic stem cells are characterized by the induced functional alterations such as increased cell-cycle capability of self-renewal and differentiation into the and proteasome activity. Detoxification enzymes and entire spectrum of blood cells. Knowledge gained from DNA repair proteins were downregulated in CML stem cell biology can also give novel insights into cancer CD34 cells, which might contribute to genetic instabil- þ research. Due to their life-long persistence and self- ity. Decreased expression of junction plakoglobulin and renewal capacity, stem cells have a high probability to CXC chemokine receptor 4 (CXCR-4) might facilitate the accumulate mutations that eventually result in malig- release of immature precursors from bone marrow in nant transformation. In addition, stem cells and CML. GATA-2 was upregulated in CML CD34 cells, þ malignant cells share several features such as the ability suggesting an increased self-renewal in comparison with to self-renew, similar mechanisms involved in migration normal CD34 cells. Moreover, we found upregulation þ and mechanisms for prevention of cellular aging such as of the proto-oncogene SKI and of receptors for neurome- telomerase expression (Passegue et al., 2003; Hope et al., diators such as opioid l1 receptor, GABA B receptor, 2004). adenosine A1 receptor, orexin 1 and 2 receptors and The best-studied malignant stem cell disorder is corticotropine-releasing hormone receptor. Treatment of chronic myelogenous leukemia (CML), which is char- CML progenitor cells with the selective adenosine A1 acterized by a clonal expansion of hematopoietic receptor antagonist 8-cyclopentyl-1,3-dipropylxanthine progenitor cells (Faderl et al., 1999). In 95% of patients (DPCPX) resulted in a dose-dependent significant inhibi- a specific chromosomal translocation between chromo- tion of clonogenic growth by 40% at a concentration of somes 9 and 22 is found, which results in the formation 10À5 M, which could be reversed by the equimolar addition of the BCR–ABL fusion gene. The BCR–ABL protein of the receptor agonist 2-chloro-N6-cyclopentyladenosine has an elevated tyrosine kinase activity and plays a (P 0.05). The incubation of normal progenitor cells with o central role in the pathophysiology of the disease (Daley DPCPX resulted in an inhibition of clonogenic growth to et al., 1990; Lugo et al., 1990). Different mechanisms a significantly lesser extent in comparison with CML cells have been shown to be involved in the malignant (P 0.05), suggesting that the adenosine A1 receptor is of o transformation by BCR–ABL (Deininger et al., 2000). functional relevance in CML hematopoietic progenitor The fusion oncoprotein activates mitogenic signalling cells. pathways such as the RAS and mitogen-activated Oncogene (2005) 24, 5313–5324. doi:10.1038/sj.onc.1208596; protein kinase pathway and protects CML cells from published online 4 April 2005 apoptosis. In addition, BCR–ABL induces adhesive and cytoskeletal abnormalities, which might facilitate the release of progenitors from the bone marrow (BM). *Correspondence: R Kronenwett; E-mail: [email protected] Large-scale gene expression analyses of leukemic cells Received 2 August 2004; revised 28 January 2005; accepted 4 February from patients with Ph þ CML have been used to 2005; published online 4 April 2005 investigate a genomic profile associated with response to Molecular phenotype of CML CD34 þ cells R Kronenwett et al 5314 therapy with the tyrosine kinase inhibitor imatinib from BM or acquisition of secondary chromosomal (McLean et al., 2004) and have broadened the knowl- abnormalities resulting in transformation into blast edge about alterations of gene expression in BCR–ABL- crisis. Several genes known to increase cell-cycle positive cells in comparison with normal cells (Ohmine progression and replication such as prefoldin 4, several et al., 2001; Nowicki et al., 2003). While Ohmine and co- cyclins, cell division cycle 25C, cyclin-dependent kinase workers examined CD133 þ hematopoietic progenitor 4, interferon-inducible RNA-dependent protein kinase, cells in the study of Nowicki and co-workers mono- 14-3-3 protein beta, and telomeric repeat-binding factor nuclear CML cells were used. However, mononuclear were significantly upregulated 1.4–10-fold in CML cells are a heterogenous cell population consisting of a CD34 þ cells. On the other hand, expression of the broad spectrum of partially and completely differen- cell-cycle inhibitors cyclin-dependent kinase inhibitors tiated cell types. Therefore, the expression pattern might 1A and 2D were 2–3-fold greater in normal CD34 þ only reflect the different proportions of the leukocyte cells than in CML CD34 þ cells. subtypes in normal and CML mononuclear cells Looking at apoptosis-related genes, we found the (Szaniszlo et al., 2004). In order to compare similar cell antiapoptotic gene defender against cell death 1 to be subsets, to analyse a more homogenous cell population 2.6-fold upregulated in CML and the proapoptotic and to examine cells which are closer to the cell of origin genes caspase 2 and death-associated protein kinase 1 to in CML, we examined highly enriched CD34 þ hema- be downregulated. topoietic stem and progenitor cells. By means of cDNA Examining genes involved in adhesion, we found a array technology, hierarchical cluster analysis, quanti- heterogenous expression pattern. The two subunits of tative real-time RT–PCR, and flow cytometry, we the a4b1 integrin (VLA-4) were three-fold upregulated, identified a distinct gene expression pattern of malignant whereas the b3 integrin subunit, intercellular adhesion CD34 þ cells from patients with untreated newly molecule 1, sialophorin, and junction plakoglobin (JUP) diagnosed CML. We found numerous differentially were downregulated between 1.7- and 4.3-fold. expressed genes and demonstrated the functional role At present, the mechanisms underlying the transfor- of the adenosine A1 receptor for the clonogenic growth mation of BCR–ABL-positive cells into blast crisis are of leukemic progenitor cells. not known. A genetic instability is presumed as the basis for accumulation of mutations. Our expression data support this view since several detoxification enzymes and genes involved in DNA repair such as microsomal Results glutathione S-transferase, glutathione peroxidase 1, glutathione reductase, cytosolic superoxide dismutase, Gene expression profiles of BM-CD34 þ cells in CML mutL homolog 1, RAD23A, ATP-dependent DNA Gene expression profiles of highly enriched CD34 þ ligase 1, and the DNA excision repair proteins ERCC1, cells from the BM of five patients with untreated CML ERCC3, and ERCC5 were significantly 2–4.4-fold lower in chronic phase were compared with those of 10 healthy expressed in CML CD34 þ cells compared with normal donors (HD) using cDNA arrays covering 1185 defined CD34 þ cells. genes. A total of 158 genes of different functional groups We also found differential expression of genes which were significantly differentially expressed in CML might be suitable for targeted therapy. For example, CD34 þ cells (Po0.01), with 55 of them upregulated three proteasome subunits were 3–4-fold overexpressed and 100 downregulated in CML (Figure 1). Complete in CML, whereas the proteasome inhibitor subunit 1 expression data are shown in the Supplementary table was 1.4-fold downregulated. available at Oncogene’s website. Moreover, we found a distinct expression pattern of First, we looked at genes which are known to be growth factors, cytokines, chemokines and their recep- involved in the BCR–ABL signal transduction pathway. tors in CML CD34 þ cells. Only one was upregulated in Several proteins downstream of BCR–ABL such as CML (small inducible cytokine subfamily E1, 3.6-fold), members of the RAS family and JAK2 were 2–7-fold whereas 17 genes were significantly lower expressed in upregulated in CML. On the other hand, the serine– CML, including the cytokines and growth factors threonine kinase AKT, which is part of the BCR–ABL- macrophage colony-stimulating factor, platelet-derived activated PI3 kinase pathway, showed a 2.4-fold growth factor, and interleukin 10, as well as the reduced expression in CML CD34 þ cells. neuromedin B receptor, the granulocyte colony-stimu- In addition, the expression pattern reflected and lating factor receptor, and the CXC chemokine receptor explained the functional characteristics of CML pro- 4 (CXCR-4). genitor cells on a molecular level, such as increased Looking at genes encoding proteins involved in proliferation activity, release of immature precursors transcription, zinc-finger proteins 9, 148, and 162,

Figure 1 Selection of differentially expressed genes. Results from ﬁve samples of CML CD34 þ cells (CML1-5, Table 1) and 10 samples of normal CD34 þ cells (HD1-10) are shown. Relative gene expression following normalization is color coded as indicated at the bottom. The medians of the relative expression levels of CML and normal samples, the ratio of those two medians (CML/HD), as well as P-values from Mann–Whitney u-test are indicated for each gene. GeneBank numbers and gene symbols according to HUGO nomenclature (Wain et al., 2002) are given

Oncogene Molecular phenotype of CML CD34 þ cells R Kronenwett et al 5315

Oncogene Molecular phenotype of CML CD34 þ cells R Kronenwett et al 5316 positive cofactor 4, hypoxia-inducible factor 1, and average linkage clustering algorithm. We included ﬁve GATA-2 were upregulated in CML CD34 þ cells, and samples of CML CD34 þ cells from BM, three samples immediate early protein, DNA-binding protein TAX- of CML CD34 þ cells from PB, 10 samples of normal REB302, E2F transcription factor 1, interferon regulatory factor, and myeloid cell nuclear differentiation antigen were downregulated in comparison to normal CD34 þ cells. We also examined proto-oncogenes and found most of them to be downregulated in CML CD34 þ cells. The proto-oncogene SKI was ninefold higher expressed in CML CD34 þ cells, whereas MYB-related protein B, JUN-D, FOS, ETS variant gene 6 (TEL proto- oncogene), and nonmetastatic cells 2 protein were downregulated between 2.1- and sixfold in comparison with normal CD34 þ cells.

Corroboration by quantitative real-time PCR To corroborate some of the data obtained by cDNA arrays, expression of seven genes in CML CD34 þ cells and normal BM CD34 þ cells was examined by quantitative real-time RT–PCR using either LightCycler or TaqMan Technology. RT–PCR conﬁrmed that JUP, tripeptidyl-peptidase (CLN2), defender against cell death 1 (DAD1), splicing factor 1 (SF162), SKI, 14-3- 3 beta protein (YWHAB), and proteasome subunit a- type 3 (PSMA3) were differentially expressed (Figure 2).

Comparison of expression profiles between BM and peripheral blood (PB) CML CD34 þ cells and hierarchical cluster analysis In a recent study, we have shown significant differences between the molecular phenotypes of BM-derived and circulating normal CD34 þ cells (Steidl et al., 2002). Therefore, we asked whether CML CD34 þ cells circulating in PB have also different expression patterns in comparison with CD34 þ cells residing in the BM. In an intraindividual comparison of gene expression profiles of CD34 þ cells from the BM and PB of three patients with CML, no gene was found to be significantly differentially expressed in both compartments (data not shown). In order to evaluate the heterogeneity of the cell samples and to demonstrate a characteristic expression pattern of CML CD34 þ cells in comparison with normal CD34 þ cells either from BM or from PB, a hierarchical cluster analysis was performed using an

Figure 2 Corroboration by quantitative real-time RT–PCR. Real- time RT–PCR curves of ﬁve genes assessed by LightCycler technology (a) and two genes assessed by Taqman technology (b) as well as of the reference gene GAPDH of a representative experiment performed in duplicate are shown. The differences of the CP or CTvalues of target and GAPDH ( DCP, DCT) are indicated. The smaller the DCP or DCT, the higher the relative expression level of the target mRNA. (c) Summary of differential expression of the seven genes corroborated by RT–PCR in comparison with cDNA array analysis. Mean values and standard deviations from three samples in duplicate assessed by real-time RT–PCR are indicated

Oncogene Molecular phenotype of CML CD34 þ cells R Kronenwett et al 5317 BM, and 12 samples of normal PB CD34 þ cells into the specific agonists or antagonists of the respective analysis. We identified three distinguishable array receptor. Incubation of cells with the agonist DAMGO clusters: (1) a homogenous group of CML CD34 þ or the antagonist Naloxone of the opioid m1 receptor cells, independent of whether they were obtained from did not influence clonogenic growth of malignant BM or PB, (2) the group of normal BM CD34 þ cells, progenitor cells (Figure 6a). On the other hand, and (3) the group of normal PB CD34 þ cells (Figure 3). incubation of cells with the specific adenosine A1 The dendrogram shows that CML CD34 þ cells were receptor antagonist DPCPX resulted in significant more similar to normal CD34 þ cells from BM than dose-dependent inhibitions of growth of CFU-GM by from PB. 40 and 47%, as well as of BFU-E by 25 and 37% using concentrations of 10À5 and 10À4 M, respectively À5 Expression of G protein-coupled neurobiological receptors (Figure 6b). The effect of the antagonist at 10 M could be significantly reversed by the equimolar addition of Recently, we have shown that genes primarily expressed the selective agonist CCPA (Po0.01 for CFU-GM, in the nervous system may also be expressed in normal Po0.05 for BFU-E). The incubation with agonist alone human CD34 þ cells (Steidl et al., 2004). This finding had no effect. prompted us to examine our CML CD34 þ cells with In comparison with leukemic progenitor cells, clono- regard to the expression of genes involved in neurobio- genic growth of normal hematopoietic progenitor cells logical functions. Out of the 56 neurobiological genes was less inhibited as incubation with the adenosine A1 covered by our cDNA array, we found the opioid m1 receptor antagonist at a concentration of 10À5 M resulted and the adenosine A1 receptors, as well as the chloride in a reduction of CFU-GM formation of 26%, whereas channel 1A, to be 3–7-fold upregulated, and neurotro- BFU-E formation was not changed (Figure 6c). The phin 3 to be twofold downregulated in CML CD34 þ reversal of the antagonist-induced effect on normal cells. CFU-GM by equimolar addition of agonist was not Next, we focused on the expression of G protein- significant (P ¼ 0.12). Therefore, the effect of DPCPX coupled receptors for neuromediators on malignant on the clonogenic growth of normal CFU-GM might be CD34 þ hematopoietic progenitor cells in CML using unspecific. Comparing the degree of inhibition of indirect immunofluorescence analysis. We corroborated DPCPX-treated (10À5 M) normal and CML progenitor the differential mRNA expression of the adenosine A1 cells by t-test for unpaired samples, we found a and the opioid m1 receptors also on the protein level, significant higher inhibition of CFU-GM as well as of showing a 1.9- and 1.8-fold higher expression of those BFU-E growth in CML (Po0.05). receptors in CML, respectively (Po0.05) (Figures 4 and 5). Moreover, the GABA B receptor, the serotonine 1F receptor as well as the receptors for the hypothalamic Discussion peptides corticotropine-releasing hormone (CRH) and orexin were 1.8–2.2-fold upregulated in CD34 þ cells In this study, we provide for the first time gene from patients with CML in chronic phase in comparison expression profiles of highly enriched CD34 þ hemato- to normal CD34 þ cells, as shown by flow cytometry on poietic stem and progenitor cells from patients with single cells (Po0.05) (Figures 4 and 5). Though the CML in chronic phase and show a characteristic gene adenosine A2B receptor and the opioid k1 receptor were expression pattern in comparison to normal CD34 þ expressed in CML CD34 þ cells, a significant higher cells. In previous studies, gene expression analyses expression in comparison to normal CD34 þ cells could were performed using BCR–ABL-positive mononuclear not be demonstrated. cells of patients with CML in chronic phase (Nowicki Since we have previously shown that the expression of et al., 2003). An important limitation of this approach those receptors on normal CD34 þ cells depends on the was that a very heterogenous population of CML coexpression of CD38 (Steidl et al., 2004), we assessed cells was used, consisting of a broad spectrum of the proportions of CD34 þ cells with low or high CD38 different precursors. This cell population was compared expression. Comparing CML CD34 þ cells with normal with fully differentiated normal control cells. In CD34 þ cells, we found a similar proportion of CD38 another study, CD133 was used to enrich hematopoietic low expressing cells in both groups (4.26% (mean) vs progenitor cells from patients with CML for gene 4.33% (mean)), indicating that differential expression of expression analyses (Ohmine et al., 2001). This study G protein-coupled receptors is not the result of a focused on the identification of molecular events different composition of the CD34 þ cell population associated with disease progression. In our study, we with regard to maturation state. used highly enriched CD34 þ cells which still represent a heterogenous population containing stem and Clonogenic growth following modulation of receptor progenitor cells. However, CD34 þ cells are closer to activity the cell of origin of the leukemic clone than mononuclear cells. Moreover, we showed by two-color For assessment of the functional role of opioid m1 and immunofluorescence analysis of CD38 coexpression adenosine A1 receptors, mononuclear cells from the BM that the cellular composition with regard to maturation of patients with CML were subjected to clonogenic state was similar in the leukemic and control CD34 þ methylcellulose assays in the presence or absence of cell samples.

Oncogene Molecular phenotype of CML CD34 þ cells R Kronenwett et al 5318 Comparing gene expression patterns between expressed genes. Differential expression of 16 genes was CMLÀCD34 þ cells with those of normal CD34 þ corroborated either by quantitative real-time RT–PCR cells from BM, we found 158 signiﬁcantly differentially or ﬂow cytometry. The gene expression pattern of CML

Oncogene Molecular phenotype of CML CD34 þ cells R Kronenwett et al 5319 CD34 þ cells reﬂected and explained functional altera- genes involved in cell-cycle progression were signiﬁ- tions which are supposed to be involved in BCR–ABL- cantly upregulated in CML, whereas out of the group of induced malignant transformation. For example, several apoptosis-related genes only few were differentially

Figure 4 Protein expression of G protein-coupled receptors on normal and malignant CD34 þ cells. Highly enriched CD34 þ cells from the BM of a healthy volunteer and of a patient with CML were assessed by two-color immunoﬂuorescence analysis using a FITC- conjugated anti-CD34 antibody and receptor-speciﬁc antibodies detected by a rhodamine-conjugated secondary antibody. The cutoff level for receptor expression was set according to the isotype control and is indicated by a line. The proportion of positive cells is given in each histogram. A representative experiment is shown

Figure 3 Hierarchical cluster analysis. The 30 samples subjected to cluster analysis were derived from CML CD34 þ cells from BM (CML1-5), CML CD34 þ cells from PB (CML6-8), normal CD34 þ cells from BM (BM1-10) or normal CD34 þ cells from PB (PB1- 12). Cluster analysis was performed on the basis of 60 genes which were at least 1.4-fold differentially expressed comparing CML CD34 þ cells from BM either with normal CD34 þ cells from BM or with normal CD34 þ cells from PB. Gene symbols are given according to HUGO nomenclature. Data are displayed by a color code. Red ﬁelds indicate higher values than the median, green ﬁelds indicate lower values than the median. The dendrogram visualizes the degree of similarity of the different subgroups by the branch length

Oncogene Molecular phenotype of CML CD34 þ cells R Kronenwett et al 5320 expressed. Our ﬁndings support the view that enhanced is responsible for the pronounced expansion of leukemic cell-cycle activation of CML CD34 þ cells (Cortez et al., progenitors (Gora-Tybor et al., 1998; Traycoff et al., 1997) rather than decreased apoptosis (Bedi et al., 1994) 1998). An interesting ﬁnding is the upregulation of proteasome subunits in CML. Recent reports have shown that BCR–ABL induces a proteasome-mediated degradation of inhibitory proteins in Ph-positive aggressive leuke- mias (Dai et al., 1998; Jonuleit et al., 2000). The

Figure 5 Differential expression of G protein-coupled receptors in CML CD34 þ cells. Summary of protein expression data of G protein-coupled receptors on BMÀCD34 þ cells from three healthy volunteers and three patients with CML is shown. Cells were stained as described in Figure 4. For each sample, mean fluorescence intensities (MFI) of cells stained with isotype-control antibody as well as of cells stained with the respective specific antibody were assessed. MFI of the sample stained with specific antibody was divided by the MFI of the respective isotype control antibody-stained sample resulting in the relative expression level (fold increase over control antibody). Mean values and standard deviations of relative expression levels of normal (white bars) and CML (gray bars) CD34 þ cells are displayed. Statistically significant higher expression in CML is indicated by asterisk (Po0.05)

Figure 6 Clonogenic growth of CML and normal progenitor cells following receptor modulation. Mononuclear cells of BM of patients with CML and healthy volunteers were examined by semisolid methylcellulose clonogenic assays in the presence or absence of receptor agonists and/or antagonists. BFU-E (gray) and CFU-GM (white bars) were counted after two weeks. Mean values and standard deviations are shown. (a) MNC from three patients with CML were examined using the opioid m1 receptor agonist DAMGO and/or antagonist naloxone. (b) MNC of patients with CML were incubated with the adenosine A1 receptor agonist CCPA and/or antagonist DPCPX at concentrations of 10À6 M (n ¼ 5), 10À5 M (n ¼ 5) or 10À4 M (n ¼ 2). The degree of inhibition was referred to the samples incubated with DMSO alone, which was the diluent for CCPA and DPCPX. Combined administration of CCPA and DPCPX at a concentration of 10À4 M as well as the respective amount of diluent alone resulted in a complete inhibition of clonogenic growth due to the high concentration of DMSO in the culture medium (20 ml/ml). Statistically significant inhibition of DPCPX referring to the DMSO control is indicated by a cross (Po0.05) or asterisk (Po0.001). Statistically significant reversal of DPCPX effects (10À5 M) on CFU-GM and BFU-E growth by equimolar addition of CCPA is indicated, with the respective P- value by brackets. (c) MNC of three healthy volunteers were incubated with the adenosine A1 receptor agonist and/or antagonist at concentrations of 10À6 and 10À5 M. Statistically significant inhibition of DPCPX referring to the DMSO control is indicated by a cross (Po0.05). Reversal of DPCPX-induced inhibition of CFU-GM growth could not be shown with sufficient statistical significance. P-value is indicated

Oncogene Molecular phenotype of CML CD34 þ cells R Kronenwett et al 5321 overexpression of proteasome subunits in CML suggests be responsible for increased proliferation of CML that proteasome-mediated degradation also plays a role progenitor cells. in chronic phase CML, which might have therapeutic In our study, we show for the first time that the SKI implications. This is supported by the finding that proto-oncogene is upregulated in CML CD34 þ cells, proteasome inhibitors induced apoptosis in BCR–ABL- suggesting a pathophysiological role in CML. SKI positive cells and had greater inhibitory effects on regulates terminal differentiation of skeletal muscle clonogenic growth of CML progenitor cells than normal (Namciu et al., 1995), and a viral homolog of SKI CD34 þ cells (Soligo et al., 2001; Yu et al., 2003). induced the proliferation of myeloid–erythroid multi- One reason for abnormal proliferation and egress of potential progenitor cells from avian BM (Dahl et al., progenitor cells and partially differentiated myeloid 1998). The SKI-induced effects in hematopoietic pro- precursor cells from BM into the blood might be a genitor cells seem to be caused in part by repression of defect of adhesive properties of CML cells (Gordon the retinoic acid receptor (RAR) signaling pathway, et al., 1987; Verfaillie et al., 1997). Interestingly, we since high concentrations of RARa-specific ligands showed that JUP is downregulated in CML. JUP is part could revert transformation of avian progenitor cells of the cellular cadherin–catenin complex and plays a by SKI. Therefore, SKI or RARa might be interesting role in cellular adhesion (Ben Ze’ev and Geiger, 1998). candidates for targeted therapy in CML. Its downregulation could be responsible for the adhesive The mechanisms underlying transformation of abnormalities of CML cells. Another interesting finding chronic phase CML into blast crisis are still poorly is decreased expression of the chemokine receptor understood. A presumed BCR–ABL-induced genetic CXCR-4 on CML CD34 þ cells. CXCR-4 plays a role instability might result in somatic mutations such as loss in mobilization and homing of normal CD34 þ cells of p53 function, which leads to blastic transformation (Mohle et al., 1998). Circulating CD34 þ cells expressed (Honda et al., 2000). Supporting this view, we found CXCR-4 at a lower level than sedentary CD34 þ cells several genes encoding detoxification enzymes and (Steidl et al., 2002), and antagonists against the proteins responsible for DNA repair to be down- chemokine receptor resulted in mobilization of hemato- regulated in chronic phase CML CD34 þ cells in poietic progenitor cells (Liles et al., 2003; Devine et al., comparison to normal CD34 þ cells. Those genes 2004). Therefore, the reduced CXCR-4 expression in include glutathione S-transferases, reduced expression CML CD34 þ cells might contribute to the release of of which contributes to an increased genetic instability malignant progenitor cells from BM. Surprisingly, we in AML, for example (Haase et al., 2002). Together with were unable to detect differentially expressed genes a decreased expression of genes involved in DNA repair, when comparing CML CD34 þ cells from BM with such as DNA excision repair proteins, this might those from PB, although we and others had observed facilitate further mutations in CML CD34 þ cells, distinct molecular phenotypes when examining normal which are often encountered after cytotoxic therapy CD34 þ cells from BM and PB (Graf et al., 2001; Steidl (Izumi et al., 1996) or in advanced phases of CML et al., 2002; Ng et al., 2004). In normal BM CD34 þ (Deininger et al., 2000). cells, a significantly higher expression of genes governing Further, we were interested in the expression pattern cell-cycle progression was found in comparison to PB of genes characteristic for hematopoietic stem cells. CD34 þ cells. Therefore, the physiological switch from Genes playing a role in self-renewal and differentiation a more proliferating to a quiescent phenotype, which covered by our array were GATA-2, GATA-3, and C/ was found during egress of normal CD34 þ cells from EBPa (Zhu and Emerson, 2002; Passegue et al., 2003; BM to PB might be disturbed in chronic phase CML. Steidl et al., 2003). GATA-2, an essential transcription Oncogenes and tumor suppressor genes are another factor for self-renewal and maintainance of stem cells in interesting group of genes with respect to malignant an undifferentiated stage, was 1.8-fold upregulated in transformation. We found a significant downregulation CML CD34 þ cells. On the other hand, GATA-3 and of ETS variant gene 6 (ETV6; TEL oncogene), JUN-D, C/EBPa, transcription factors governing lymphoid and and FOS in CML progenitor cells. ETV6 is essential for myeloid lineage commitment decisions, respectively, the establishment of hematopoiesis of all lineages in the were not expressed at all in CML CD34 þ cells. These BM (Wang et al., 1998). Moreover, ETV6 is involved in findings suggest that malignant hematopoietic stem cells chromosomal translocations associated with hematolo- in CML have a greater ability to self-renew in gical malignancies as, for example, the ETV6–AML1 comparison with normal CD34 þ cells. gene fusion in t(12;21)-positive childhood acute lym- Finally, in this study, we show for the first time that phoblastic leukemia (ALL) (Wlodarska et al., 1996). leukemic hematopoietic progenitor cells expressed sev- Therefore, downregulation of ETV6 in hematopoietic eral G protein-coupled receptors which were primarily progenitor cells could be involved in the molecular assigned to the nervous system, such as GABA B pathogenesis of CML. JUN-D antagonizes the positive receptor, opioid m1 and k1 receptors, serotonine 1F regulatory effects of c-JUN on cell cycle and is expressed receptor, adenosine A1 and A2B receptors, and recep- preferentially in quiescent cells (Musti et al., 1996; tors of the hypothalamic peptides corticotropin-releas- Liebermann and Hoffman, 1998). In the same line, FOS ing hormone (CRH) and orexins. In a previous study, plays a role in growth suppression and apoptosis in our group has shown that normal primary human many cell types (Liebermann and Hoffman, 1998). CD34 þ cells express functionally active receptors of Therefore, downregulation of JUN-D and FOS might neuromediators (Steidl et al., 2004). However, the

Oncogene Molecular phenotype of CML CD34 þ cells R Kronenwett et al 5322 functional relevance of those receptors for normal and upregulation of the adenosine A1 receptor in leukemic malignant hematopoiesis is not known so far. Since CD34 þ cells, although rather mild, is of functional seven of the nine assessed receptors are significantly significance in CML. The adenosine A1 receptor might upregulated in leukemic CD34 þ cells in comparison therefore be a target for therapeutic approaches with with normal CD34 þ cells, one might speculate that specific receptor antagonists. they play a pathophysiological role in CML. As assessed In conclusion, our data provide a gene expression by two-color immunofluorescence analysis, the propor- profile of primary CD34 þ hematopoietic stem and tion of more immature CD38-negative cells was similar progenitor cells of patients with CML in chronic phase. in CML and normal CD34 þ cells. Therefore, differ- Gene expression patterns reflected and explained the ential expression of G protein-coupled receptors due to functional properties of leukemic cells on a molecular different composition of the CD34 þ cell populations is level. Furthermore, we identified genes such as the SKI unlikely. We assessed the functional role of two of those proto-oncogene or receptors of neuromediators which G protein-coupled receptors with respect to clonogenic might be of functional relevance in CML. growth by treating cells with specific agonists and antagonists. Incubation of CML mononuclear cells with DAMGO, a selective agonist of the opioid m1 receptor, Materials and methods or with the antagonist Naloxone had no effect on the clonogenic growth of lineage-committed progenitor Cells cells. This is surprising since in a previous study an increased proliferation of granulocyte-macrophage, er- After informed consent, BM mononuclear cells and PB ythroid and multipotential progenitor cells in BM and mononuclear cells were obtained by density centrifugation from patients with untreated newly diagnosed Ph þ CML in spleen was found in opioid m1 receptor-deficient mice, chronic phase. Patients’ characteristics are given in Table 1. All suggesting a negative regulatory influence on normal patients had 100% Philadelphia chromosome-positive meta- hematopoiesis (Tian et al., 1997). Incubation of CML phases in standard BM cytogenetics. Normal BM cells from 10 mononuclear cells with a specific antagonist of the healthy volunteers and PB mononuclear cells from 12 healthy adenosine A1 receptor resulted in a significant dose- volunteers were obtained as described previously (Steidl et al., dependent inhibition of clonogenic growth of both, 2002). CFU-GM and BFU-E, suggesting a role in proliferation CD34 þ cells were positively selected using the midiMACS and differentiation of myelomonocytic as well as immunomagnetic separation system (Miltenyi Biotec, Bergisch erythoid lineage-committed progenitor cells in CML. Gladbach, Germany) as described previously (Steidl et al., This is in line with recent data showing that extracellular 2002). Purities of CML CD34 þ cells are given in Table 1. Purities of normal CD34 cells ranged between 90 and 99%. adenosine enhances cell cycling of hematopoietic pro- þ genitors (Pospisil et al., 2001). The incubation of normal mononuclear cells with the adenosine A1 receptor RNA isolation, reverse transcription, and hybridization to nylon antagonist resulted in an inhibition of CFU-GM cDNA arrays formation to a significantly lesser extent in comparison Isolation of total RNA from 2 Â 105 to 2.6 Â 107. CD34 þ cells with CML mononuclear cells and did not affect growth was performed with the RNeasy Mini Kit (Qiagen of normal BFU-E at all. This suggests that the AG, Hilden, Germany) according to the manufacturer’s

Table 1 Patients’ characteristics No. Age Leuko (/nl) Thrombo (/nl) Hb (g/dl) BCR–ABL/G6PDH ratio (%) Purity (%)

Gene expression analyses 1 70 33 725 14.2 33.8 99.9 2 25 126 200 12.6 18.4 96.2 3 61 78 620 12.9 10.8 97.0 4 43 58 557 12.5 7.4 99.4 5 60 33 1169 13.2 22.0 94.0

Flow cytometry 6 39 129 261 9.4 9.5 98.9 7 58 48 491 13.3 14.7 98.6 8 27 164 842 12.1 11.2 99.6

Clonogenic assays 9 41 180 476 12.1 5.4 n.p. 10 51 28 362 12.4 4.4 n.p. 11 54 206 331 10.1 25.8 n.p. 12 39 395 344 9.8 2.1 n.p. 13 52 116 1013 11.6 14.9 n.p.

Age, PB counts (leukocytes, thrombocytes, hemoglobin), BCR–ABL transcript levels (BCR–ABL/G6PDH ratio) and purity of enriched CD34+ cells is given for each patient included in this study. BCR–ABL/G6PDH ratio was assessed as previously described (Neumann et al., 2003). For clonogenic assays, no CD34+ cell selection was performed (n.p.)

Oncogene Molecular phenotype of CML CD34 þ cells R Kronenwett et al 5323 instructions. Atlas Human 1.2 I arrays (BD Biosciences of the manufacturer of the Assays-on-demand using an ABI Clontech, Heidelberg, Germany) were used for hybridization PRISMs 7900HTSequence Detector. Thefollowing Assay- experiments. Between 600 ng and 5 mg RNA were reverse- on-Demand Gene Expression Products (Applera) were used: transcribed and radioactively labeled as described previously proteasome subunit a-type 3 (Hs00160558_m1), splicing factor (Steidl et al., 2002). 1 (Hs00190309_m1), and GAPDH (Hs99999905_m1). Ct values were calculated by the ABI PRISM software and Quantification, normalization, and statistical analysis relative gene expression levels were expressed as the difference of CTvalues of target gene and GAPDH ( DCT). Phosphorimager screens were scanned using a Phosphorima- ger (Fuji FLA-3000, FujiFilm) controlled by the BAS-Reader 3.01. Software (Raytest, Straubenhardt, Germany). Dot Immunofluorescence staining and flow cytometry intensities were measured with the Aida Image Analyzer Immunomagnetically selected CD34 þ cells were stained with Software (Raytest, Straubenhardt, Germany). After back- a fluorescein isothiocyanate-conjugated (FITC) monoclonal ground subtraction, raw data were normalized using the anti-CD34 antibody (clone 8G12, Becton Dickinson, Heidel- median intensity of the 20% highest expressed genes and berg, Germany) as described (Steidl et al., 2002). Analysis of expressed as the ratio (dot intensity)/(median intensity). To CD38 coexpression and detection of opioid m1 and k1 compare the CD34 þ cell samples of patients with CML with receptors, GABA B receptor, serotonin (5-HT) 1F receptor, the samples of HD, for each gene the median values of the adenosine A1 and A2B receptors, CRH receptor, and orexin CML and the HD group were calculated and the CML median receptors 1 and 2 by indirect immunofluorescence was was divided by the HD median obtaining the ratio CML/HD. performed as described (Steidl et al., 2004). For calculation of significant differential expression, the Mann–Whitney u-test was used. P-values smaller than 0.01 were accepted as statistically significant. Semisolid clonogenic assays Hierarchical cluster analysis was performed by the ‘hclust’ MNC from the BM of untreated patients with CML in chronic function of the statistical scripting language ‘R’ integrated into phase were seeded in semisolid ready-to-use methylcellulose the Bioconductor project (http://www.bioconductor.org/) growth medium containing stem cell factor, granulocyte- using an average linkage clustering algorithm. macrophage colony-stimulating factor, granulocyte colony- stimulating factor, interleukin-3, interleukin-6, erythropoietin Quantitative real-time reverse transcription polymerase chain (MethoCult H4436, StemCell Technology, Vancouver, Cana- reaction (RT–PCR) da) at concentrations ranging between 1 Â 105 and 1 Â 106 cells/ Corroboration of RNA expression was performed by real-time ml with agonist DAMGO (Sigma, Taufkirchen, Germany) or RT–PCR using either the LightCyclers technology (Roche antagonist Naloxone (DuPont, Bad Homburg, Germany) of Molecular Biochemicals, Mannheim, Germany) as described the opioid m1 or agonist CCPA (Sigma) or antagonist DPCPX (Steidl et al., 2002) or the ABI PRISMs 7900HTSequence (8-cyclopentyl-1,3-dipropylxanthine; Sigma) of the adenosine A1 receptors or with both (final concentrations of agonists and Detection System Instrument (Applied Biosystems, Applera À6 À5 À4 Deutschland GmbH, Darmstadt, Germany). Total RNA was antagonists: 1 Â 10 ,1Â 10 , and 1 Â 10 M) as described et al extracted using the ‘Absolute RNA Microprep’ system previously (Kronenwett ., 1998). CCPA and DPCPX were dissolved with dimethyl sulfoxide (DMSO; Sigma) before use, (Stratagene, La Jolla, CA, USA) and was reverse-transcribed À2 as described previously (Kronenwett et al., 2002). GAPDH resulting in a 10 M stock solution. Colony numbers (colony- mRNA served as external control for relative quantification. forming units granulocyte/macrophage (CFU-GM), burst- LightCycler PCR was performed using the LightCycler- forming units erythrocyte (BFU-E)) were counted after 2 FastStart DNA Master SYBR Green kit (Roche Molecular weeks. For treatment of normal hematopoietic progenitor cells Biochemicals). The following primers were used: GAPDH: 50- with CCPA and DPCPX, MNC from healthy volunteers were 0 0 used in clonogenic assays at final concentrations of agonist or TCCATGACAACTTTGGTATCG-3 ,5-GTCGCTGTTGA À6 À5 AGTCAGAGGA-30; JUP: 50-GTCAAGGCAACCATCG-30, antagonist of 1 Â 10 and 1 Â 10 M. 50-CTCCTCCATCCTCACA-30; tripeptidyl-peptidase: 50-TTC ATACCAGGAGGAAGC-30,50-CCTTGGGTTGAGAAA Statistical analysis GC-30; defender against cell death 1: 50-TAGCGGTTTGCCT For calculation of statistically significant differences between GAG-30,50-GTTGTTCTGACACACAGT-30; SKI: 50-CTTC two groups in quantitative real-time RT–PCR, flow cytometry, CAATAAGAGCCTG-30,50-ATGAGGTAAAGGACGG-30; and clonogenic assays, the t-test for unpaired or paired 14-3-3 beta protein: 50-GACAACAAACAAACCAC-30,50- samples was used. A P-value 0.05 was considered to be GTCCCTAAGTAACTGC-30. Primers were designed using o statistically significant. the ‘LightCycler Probe Design’ software 1.0 (Roche Molecular Biochemicals). The crossing points (CP) of real-time PCR curves were determined by the LightCycler 3.5 software using Acknowledgements the second derivative maximum method. Relative gene We thank Baerbel Junge, Anke Boeckmann, Hildegard expression levels were expressed as difference of CP values of Gaussmann, Sabrina Pechtel, Monika Pooten, and Elke target gene and GAPDH (DCP). Rosenbaum-Koenig for expert technical support. This work ABI PRISMs PCR was performed in a MicroAmp Optical was financially supported by Leukaemie Liga e.V. Duesseldorf 96-well Reaction Plate (Applera) according to the instructions and Moonlife e.V. Karlsruhe.

References

Bedi A, Zehnbauer BA, Barber JP, Sharkis SJ and Jones RJ. Ben Ze’ev A and Geiger B. (1998). Curr. Opin. Cell Biol., 10, (1994). Blood, 83, 2038–2044. 629–639.

Oncogene Molecular phenotype of CML CD34 þ cells R Kronenwett et al 5324 Cortez D, Reuther G and Pendergast AM. (1997). Oncogene, Namciu S, Lyons GE, Micales BK, Heyman HC, Colmenares 15, 2333–2342. C and Stavnezer E. (1995). Dev. Dyn., 204, 291–300. Dahl R, Kieslinger M, Beug H and Hayman MJ. (1998). Proc. Neumann F, Herold C, Hildebrandt B, Kobbe G, Aivado M, Natl. Acad. Sci. USA, 95, 11187–11192. Rong A, Free M, Ro¨ ssig R, Fenk R, Schneider P, Dai Z, Quackenbush RC, Courtney KD, Grove M, Cortez D, Gattermann N, Royer-Pokora B, Haas R and Kronenwett Reuther GW and Pendergast AM. (1998). Genes Dev., 12, R. (2003). Eur. J. Haematol., 70, 1–10. 1415–1424. Ng YY, van Kessel B, Lokhorst HM, Baert MR, van den Burg Daley GQ, Van Etten RA and Baltimore D. (1990). Science, CM, Bloem AC and Staal FJ. (2004). J. Leukoc. Biol., 75, 247, 824–830. 314–323. Deininger MW, Goldman JM and Melo JV. (2000). Blood, 96, Nowicki MO, Pawlowski P, Fischer T, Hess G, Pawlowski T 3343–3356. and Skorski T. (2003). Oncogene, 22, 3952–3963. Devine SM, Flomenberg N, Vesole DH, Liesveld J, Weisdorf Ohmine K, Ota J, Ueda M, Ueno S, Yoshida K, Yamashita Y, D, Badel K, Calandra G and DiPersio JF. (2004). J. Clin. Kirito K, Imagawa S, Nakamura Y, Saito K, Akutsu M, Oncol., 22, 1095–1102. Mitani K, Kano Y, Komatsu N, Ozawa K and Mano H. Faderl S, Talpaz M, Estrov Z, O’Brien S, Kurzrock R and (2001). Oncogene, 20, 8249–8257. Kantarjian HM. (1999). N. Engl. J. Med., 341, 164–172. Passegue E, Jamieson CH, Ailles LE and Weissman IL. (2003). Gora-Tybor J, Deininger MW, Goldman JM and Melo JV. Proc. Natl. Acad. Sci. USA, 100 (Suppl 1), 11842–11849. (1998). Br. J. Haematol., 103, 716–720. Pospisil M, Hofer M, Vacek A, Netikova J, Hola J, Znojil V Gordon MY, Dowding CR, Riley GP, Goldman JM and and Weiterova L. (2001). Exp. Hematol., 29, 557–562. Greaves MF. (1987). Nature, 328, 342–344. Soligo D, Servida F, Delia D, Fontanella E, Lamorte G, Graf L, Heimfeld S and Torok-Storb B. (2001). Biol. Blood Caneva L, Fumiatti R and Lambertenghi DG. (2001). Br. J. Marrow Transplant., 7, 486–494. Haematol., 113, 126–135. Haase D, Binder C, Bunger J, Fonatsch C, Streubel B, Steidl U, Bork S, Schaub S, Selbach O, Seres J, Aivado M, Schnittger S, Griesinger F, Westphal G, Schoch C, Knopp Schroeder T, Rohr UP, Fenk R, Kliszewski S, Maercker C, A, Berkovicz D, Krieger O, Wormann B, Hilgers R, Hallier Neubert P, Bornstein SR, Haas HL, Kobbe G, Tenen DG, E and Schulz T. (2002). Leuk. Res., 26, 249–254. Haas R and Kronenwett R. (2004). Blood, 104, 81–88. Honda H, Ushijima T, Wakazono K, Oda H, Tanaka Y, Steidl U, Kronenwett R, Martin S and Haas R. (2003). Vitam. Aizawa S, Ishikawa T, Yazaki Y and Hirai H. (2000). Blood, Horm., 66, 1–28. 95, 1144–1150. Steidl U, Kronenwett R, Rohr UP, Fenk R, Kliszewski S, Hope KJ, Jin L and Dick JE. (2004). Nat. Immunol., 5, Maercker C, Neubert P, Aivado M, Koch J, Modlich O, 738–743. Bojar H, Gattermann N and Haas R. (2002). Blood, 99, Izumi T, Imagawa S, Hatake K, Miura Y, Ariyama T, 2037–2044. Inazawa J and Abe T. (1996). Int. J. Hematol., 64, 73–77. Szaniszlo P, Wang N, Sinha M, Reece LM, Van Hook JW, Jonuleit T, van der KH, Miething C, Michels H, Hallek M, Luxon BA and Leary JF. (2004). Cytometry, 59A, 191–202. Duyster J and Aulitzky WE. (2000). Blood, 96, 1933–1939. Tian M, Broxmeyer HE, Fan Y, Lai Z, Zhang S, Aronica S, Kronenwett R, Graf T, Nedbal W, Weber M, Steidl U, Rohr Cooper S, Bigsby RM, Steinmetz R, Engle SJ, Mestek A, UP, Mohler Tand Haas R. (2002). Cancer Gene Ther., 9, Pollock JD, Lehman MN, Jansen HT, Ying M, Stambrook 587–596. PJ, Tischﬁeld JA and Yu L. (1997). J. Exp. Med., 185, Kronenwett R, Steidl U, Kirsch M, Sczakiel G and Haas R. 1517–1522. (1998). Blood, 91, 852–862. Traycoff CM, Halstead B, Rice S, McMahel J, Srour EF and Liebermann DA and Hoffman B. (1998). Oncogene, 17, Cornetta K. (1998). Br. J. Haematol., 102, 759–767. 3319–3329. Verfaillie CM, Hurley R, Lundell BI, Zhao C and Bhatia R. Liles WC, Broxmeyer HE, Rodger E, Wood B, Hubel K, (1997). Acta Haematol., 97, 40–52. Cooper S, Hangoc G, Bridger GJ, Henson GW, Calandra G Wain HM, Bruford EA, Lovering RC, Lush MJ, Wright MW and Dale DC. (2003). Blood, 102, 2728–2730. and Povey S. (2002). Genomics, 79, 464–470. Lugo TG, Pendergast AM, Muller AJ and Witte ON. (1990). Wang LC, Swat W, Fujiwara Y, Davidson L, Visvader J, Kuo Science, 247, 1079–1082. F, Alt FW, Gilliland DG, Golub TR and Orkin SH. (1998). McLean LA, Gathmann I, Capdeville R, Polymeropoulos MH Genes Dev., 12, 2392–2402. and Dressman M. (2004). Clin. Cancer Res., 10, 155–165. Wlodarska I, Mecucci C, Baens M, Marynen P and van den Mohle R, Bautz F, Raﬁi S, Moore MA, Brugger W and Kanz BH. (1996). Leuk. Lymphoma, 23, 287–295. L. (1998). Blood, 91, 4523–4530. Yu C, Rahmani M, Conrad D, Subler M, Dent P and Grant S. Musti AM, Treier M, Peverali FA and Bohmann D. (1996). (2003). Blood, 102, 3765–3774. Biol. Chem., 377, 619–624. Zhu J and Emerson SG. (2002). Oncogene, 21, 3295–3313.

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

Oncogene