Background
Chromosomal translocations are the most frequently occurring genetic abnormalities in leukemias and they exploit a mechanism by which normal regulatory pathways are subverted, thereby providing a proliferative advantage to a leukemic clone. Many of the chromosomal translocations involve tyrosine kinases. In most cases, translocations juxtapose a tyrosine kinase domain to another protein containing an oligomerization motif. For example, BCR-ABL is generated by t(9;22)(q34;q11) [
1], which fuses the N-terminus of BCR to the C-terminus of ABL. TEL-PDGFRB (t(5;12)(q33;p13)) [
2] and TEL-JAK2 (t(9;12)(p24;p13)) [
3,
4] fuse the N-terminus of TEL to the C-terminus of PDGFRB or JAK2, respectively. In all three fusions, the N-terminal translocation partners contain an oligomerization domain (i.e. the coiled-coil domain of BCR or the pointed domain of TEL). The oligomerization domain mediates ligand-independent activation of the kinase domains, causing factor-independence
in vitro and leukemia-like diseases
in vivo.
The kinase fusions provide a proliferative signal for the leukemic clone to expand, while additional mutations are necessary to fully elicit the leukemia phenotype. For all three fusions, the enhanced proliferative capacity is associated with activation of STAT proteins [
5‐
13], various members of MAP kinases [
14‐
18] and activation of the PI3K pathway [
19‐
22] (reviewed in [
23‐
25]). While BCR-ABL, TEL-PDGFRB and TEL-JAK2 activate overlapping signaling pathways, the relative contribution and requirement for these pathways downstream of each fusion is not clear. Given that each of these pathways can transmit signals to alter gene transcription, it is possible that their signals converge at the level of gene transcription, resulting in a similar gene expression pattern. Alternatively, they may result in distinct gene expression patterns depending on the relative significance of different pathways downstream of the three fusions.
The three fusions give rise to distinct diseases according to the WHO classification (reviewed in [
26]). BCR-ABL is predominantly described in patients with chronic myelogenous leukemia (CML). The TEL-PDGFRB translocation is commonly associated with chronic myelomonocytic leukemia (CMML), a disease characterized by a preferential expansion of monocytes [
26,
27]. TEL-JAK2 has been described in three distinct diseases: pre-B cell acute lymphoblastic leukemia (ALL), T-cell ALL and atypical CML. The differences in the clinical features of the three fusion tyrosine kinases may be, in part, due to distinct regulation of gene transcription.
In order to investigate this possibility in an unbiased manner, we used microarray analysis to compare the effects of BCR-ABL, TEL-PDGFRB and TEL-JAK2 on gene expression in Ba/F3 cells. We demonstrated that BCR-ABL, TEL-PDGFRB and TEL-JAK2 elicit distinct gene expression changes. Notably, TEL-PDGFRB, but not the other two fusions, induces the expression of Stat1 and Gbp1. Interestingly, Cd55 and Ndrg1 are induced by BCR-ABL and suppressed by TEL-PDGFRB, while Cxcl10 and Scinderin are commonly regulated by all three fusions. This study illustrates that oncogenic tyrosine kinases can trigger distinct gene regulation, which may contribute to specificity downstream of each fusion protein.
Methods
Cell lines and culture
Ba/F3 cells and Ba/F3 TEL-JAK2(5-19) cells were cultured as previously described [
10,
16,
20]. TEL-JAK2(5-19) was introduced via electroporation and individual subclones were isolated by limiting dilution.
Ba/F3 BCR-ABL cells were generated by retroviral infection of Ba/F3 cells with pMSCV BCR-ABL p210 construct. Ba/F3 TEL-PDGFRB cells were generated by electroporation. Both cell lines were maintained in RPMI-complete media supplemented with murine IL-3, Geneticin and 100 ng/ml of Imatinib. Individual G418-resistant subclones were isolated by limiting dilution, and expression of BCR-ABL and TEL-PDGFRB was confirmed by immunoprecipitation and immunoblot analysis.
Cell cycle analysis
2 × 106 cells were collected, washed once in cold PBS, fixed in ice-cold 70% ethanol and stored at -20°C. For propidium iodide staining, cells were washed twice in cold PBS and incubated in PBS containing RNAse A (Invitrogen) (1 unit per 106 cells) for 30 min at 37°C. After 30 min, 50 μg/ml of propidium iodide (Roche Boehringer Mannheim, Laval, QC) was added to the cell suspension, and cells were analyzed by FACScan. The cell cycle profile was determined by using the ModFit LT® (Verity Software House, Topsham, ME).
Imatinib dose-response assay
Ba/F3 cells stably expressing pMSCV TEL-PDGFRB or pMIGR BCR-ABL were washed three times in 10 mM HEPES (pH 7.4)/Hanks balanced salts, and incubated in RPMI-complete medium for 24 h at 37°C. After 24 h, 2 × 105 cells were resuspended in RPMI-complete media containing specific amounts of Imatinib mesylate dissolved in DMSO. Ba/F3 cells were resuspended in RPMI-complete media containing 100 pg/ml of IL-3 and increasing amounts of Imatinib. As a vehicle control, Ba/F3 were resuspended in RPMI-complete media containing DMSO only.
Ba/F3 cells were washed three times in 10 mM HEPES (pH 7.4)/Hanks balanced salts, incubated in RPMI-complete medium for 5 h at 37°C and stimulated with 2 ng/ml of rmIL-3 for 0, 8, 12, 24 h and 1 week at 37°C.
Ba/F3 BCR-ABL and Ba/F3 TEL-PDGFRB cells were washed three times and incubated in RPMI-complete media and Imatinib (1 μg/ml for Ba/F3 BCR-ABL cells and 100 ng/ml for Ba/F3 TEL-PDGFRB cells) for 5 h at 37°C. After 5 h, cells were washed three times and incubated in RPMI-complete media for 0, 8, 12, 24 h and 1 week at 37°C.
Ba/F3 TEL-JAK2 cells were washed three times and incubated in RPMI-complete media for 0, 8, 12, 24 h and 1 week at 37°C.
For all cell lines, cells were harvested at each time-point for total RNA extraction, which was performed by using QIAGEN RNAeasy MINI® (Qiagen, Mississauga, Ontario, Canada), and the integrity of the RNA samples was assessed by running a formaldehyde denaturing gel according to the manufacturer's protocol.
Oligonucleotide array experiment and data analysis
Further RNA quality assessment, sample RNA
in vitro transcription, labeling and hybridization were conducted by the Toronto Centre for Applied Genomics at the Sick Children's Hospital
http://www.tcag.ca/ according to the standard Affymetrix protocols. Hybridization was performed using the Affymetrix mouse MOE430A oligonucleotide array, which represents 13247 unique mouse gene transcripts. All arrays have been scaled to a target intensity of 150, and data was analyzed by using Affymetrix Micro Array Suite (MAS) 5.0 software to calculate expression values for each transcript and Detection p-values. The Detection
p-value was calculated by the Detection algorithm, which assigned a Present (
p < 0.04), Marginal (0.04 <
p < 0.06) or Absent (0.06 <
p) call, and reflected the confidence of the detection of the given transcript. All of the reported induced genes displayed the Detection score of Present or Marginal at the 1 week point, and all of the reported suppressed genes displayed the Detection score of Present or Marginal at the 0 h point. Ratios between the expression levels at the 0 h and all subsequent time-points were calculated.
For gene expression in Ba/F3 TEL-JAK2 cells, expression values obtained from Ba/F3 TEL-JAK2 cells at the 1 week time point were compared to the expression values obtained from Ba/F3 cells at the 0 h time point.
Clustering analysis
Gene expression values were log2 transformed, and two hundred and fifty of the most highly induced genes and two hundred and fifty of the most highly suppressed genes with a Detection score of Present or Marginal at the 1 week point (for the induced genes) or at the 0 h point (for the suppressed genes) were selected. Hierarchical clustering was performed using the Eisen Cluster and TreeView software package using the average linkage criterion [
28].
Validation by quantitative PCR (Q-PCR)
A 5 μg aliquot of total RNA was reverse transcribed using Superscript II reverse transcriptase (Invitrogen). A 10 ng aliquot of cDNA was used for each Q-PCR assay performed with the ABI Prism 7900 HT thermocycler (Applied Biosystems) using SYBR Green detection (Perkin-Elmer Applied Biosystems, Foster City, CA). Each assay was performed in triplicate using RNA samples extracted from 2-4 distinct populations of cells. The sequences of gene transcripts that correspond to the Affymetrix probe set were obtained from Genbank accession numbers listed on the Affymetrix website and primer sets corresponding to exon sequences for PCR amplifications were designed using the Primer Express software (Perkin-Elmer Applied Biosystems, Foster City, CA).
The expression of cyclophilin H was assessed in all samples, and did not alter significantly at different time-points, and therefore, was used to normalize starting cDNA concentrations, and the primers used for all Q-PCR are listed in Additional file
1: Table S1. Dissociation curves were performed as routine verification in order to check the primers for amplification of a single band, and template dilution standard curves (5-log range) were conducted with each primer set to verify a linear relationship between template concentration and Ct values (R
2 > 0.9). Absolute transcript concentrations were calculated using standard curves for each primer set. Relative ratios were then calculated for each cell type in a manner similar to the microarray ratios in order to allow relative comparison between data sets.
Antibodies, immunoprecipitation and western blotting
All reagents have been described in earlier publications [
10,
16,
20].
Ba/F3 BCR-ABL or TEL-PDGFRB cells were washed three times in 10 mM HEPES (pH 7.4)/Hanks balanced salts, and resuspended in RPMI-complete media or RPMI-complete media supplemented with IL-3 ± 1 μg/ml of Imatinib (Novartis). Lysates were prepared and analyzed by performing Western blot using the 4G10 anti-phosphotyrosine monoclonal antibody.
Discussion and conclusions
BCR-ABL, TEL-PDGFRB and TEL-JAK2 are recurrent chromosomal translocations associated with distinct forms of leukemia that differ in the target cell and in disease aggressiveness. All three fusion proteins activate similar signaling pathways involving MAP kinases, Stat proteins and PI3K in hematopoietic cell lines. Our current understanding of the cytosolic signaling pathways of these fusion proteins alone cannot explain the basis for these differences, and we reasoned that examination of gene expression changes regulated by the three fusions might provide insights in this regard.
The results from the oligonucleotide array analysis demonstrate that the three tyrosine kinase fusions trigger both overlapping and distinct gene expression changes, and these results were confirmed by Q-PCR (Table
5). In addition, we demonstrated that TEL-PDGFRB induced a large number of ISG.
Because we can attribute gene expression changes downstream of BCR-ABL and TEL-PDGFRB directly to these kinases, we chose 12 genes regulated by BCR-ABL or TEL-PDGFRB for Q-PCR validation based on the fold-change in their expression, the Detection score in the gene expression as predicted by MAS 5.0, and their cellular function. As we were interested in the genes that could potentially contribute to leukemogenesis, we selected those that modulate cell proliferation and differentiation.
Validation by Q-PCR revealed that Dok2, Isg20, Id1, Mrvi1, gp49b, Cxcl10, Scinderin and Col5a1 display an overlapping regulation among the three fusion kinases by 2-fold or greater at steady-state (Table
5). In particular, Cxcl10 and Scinderin are commonly regulated by all three fusions. Cd55 and Ndrg1 are suppressed by TEL-PDGFRB, whereas they are induced by BCR-ABL. Finally, Stat1 and Gbp1 are uniquely induced by TEL-PDGFRB.
Most studies that examine the gene expression changes associated with BCR-ABL have compared the normal hematopoietic progenitors to the CML progenitors [
32,
34‐
38]. A caveat of this approach is that it is difficult to attribute genetic changes directly to BCR-ABL as they may have arisen from secondary mutations. The use of Imatinib allowed us to link changes in gene expression directly to BCR-ABL and TEL-PDGFRB, and to examine the kinetics of gene regulation. Kinetic analysis revealed that a distinct subset of genes was regulated within the first 8 h of kinase activation (e.g. Id1, Scinderin), while others were regulated after 24 h (e.g. Stat1, Cxcl10, gp49b and Col5a1). The late changes in the expression of genes such as Stat1, Cxcl10, gp49b and Col5a1 may be attributable to the relatively small extent of changes in gene expression at the earlier time-points, and it is possible that early changes occurred but were not detectable in our system.
Many of the genes identified in our study are associated with regulation of cell migration, proliferation and differentiation, all of which are known to be perturbed during leukemogenesis. Considering the functions of the validated genes, the finding that these genes are regulated downstream of the three fusions suggest that they may play important roles during leukemia development.
Cxcl10, gp49b and Mrvi1 are membrane proteins up-regulated in our experiments. Cxcl10, is a member of the chemokine family, that plays an important role in regulating movement and retention of hematopoietic progenitor cells within the bone marrow microenvironment. Chemokines have been implicated in Imatinib resistance [
39] and as a potential mechanism for effective interferon treatment [
40]. CML cells fail to respond to at least two different chemokines: macrophage inflammatory protein (MIP)-1α and stromal-derived factor (SDF)-1 [
41‐
44]. Moreover, it was demonstrated that BCR-ABL both positively and negatively regulates SDF-1-mediated signaling suggesting that BCR-ABL autonomously regulates the cell migration and retention of hematopoietic progenitors in the marrow [
43,
45].
gp49b, also known as leukocyte immunoglobulin-like receptor, subfamily B, member 4 (Lilrb4), is a ITIM-containing, Ig-like receptor expressed on mast cells that plays an inhibitory role in IgE- or cytokine-dependent mast cell activation and inflammation [
46‐
48]. gp49b was found to be up-regulated at the mRNA level by G-CSF stimulation in multiple myeloid cell lines [
49] and its related member, LILRB3 has been reported to be highly induced in CML-chronic phase samples in microarray experiments [
32]. While we observed a transient induction of gp49b around the 8 h time-point in some instances (Figure
5), this transient induction was not sustained at 24 h. Given that gp49b is significantly elevated at 1 week, our findings suggest that the induction of gp49b may be subsequent to changes that occur early in response to BCR-ABL activation.
Col5a1 is a member of Type V collagen family, which mediates cell adhesion by binding preferentially to α
2β
1 integrin [
50]. Although no direct evidence exists between Type V collagens and leukemia, adhesion receptors play a critical role during hematopoiesis as they relay the external cues to the cells, and CML cells are defective in cell adhesion to the bone marrow stroma and extracellular matrix. Moreover, Col1a1 and Col4a4, have been previously reported to be highly suppressed in CML chronic phase samples in cDNA microarray experiments [
32,
51‐
56]. Col5a1 is a HIF-1α target gene, and the expression of HIF-1α has been shown to be a positive prognostic factor in lymphoma [
57]. In our study, Col5a1 was significantly suppressed at steady state by both BCR-ABL and TEL-PDGFRB, suggesting that its suppression may contribute to leukemia caused by these two fusions.
Several signal transducers were regulated by activated tyrosine kinases in our experiments including Dok2 and Scinderin. Dok2 was consistently elevated downstream of BCR-ABL and TEL-PDGFRB. Dok1 and Dok2 exhibit increased tyrosine phosphorylation in p210
BCR-ABL-expressing cell lines [
58] and tyrosine phosphorylation of Dok2 was significantly down regulated in Mo7e/p210
BCR-ABL cells treated with Imatinib [
59]. Dok2 mediates cytoskeletal activity by its direct interaction with c-Abl, resulting in increased c-Abl tyrosine phosphorylation and kinase activity [
60]. In contrast to our study, Nowicki et al. observed a strong down-regulation of the Dok2 transcript in the bone marrow cells of patients with CML at blast crisis [
32]. The study by Nowicki et al., however, has a few caveats, which may explain the discrepancy between our results. Firstly, the finding was made by cDNA microarray analysis, and the result was not validated by a quantitative assay. Secondly, the authors compared a highly heterogeneous population of CML samples from peripheral blood mononuclear cells to fully differentiated normal control cells. On the other hand, the discrepancy between these reports may reflect differences in the experimental systems employed in each of these studies. In addition, two studies demonstrated that a compound Dok1/Dok2 knockout mouse developed a CML-like disease [
61,
62]. We observed no difference in Dok1 expression in Ba/F3 BCR-ABL cells in our microarray experiments. Additionally, the role of Dok2 downstream of TEL-PDGFRB remains to be explored.
Scinderin (adseverin) was highly suppressed by all three fusions at steady state, and this occurred as early as 8 h after BCR-ABL or TEL-PDGFRB activation. Scinderin is a Ca
2+-dependent actin severing protein, expressed in normal megakaryocytes and platelets that controls actin filament length [
63]. The role of Scinderin is crucial in regulating cortical F-actin to mediate secretion including the release of platelets from mature megakaryocytes. Proteomic analysis of Ba/F3 BCR-ABL and Ba/F3 TEL-PDGFRB cells revealed that Scinderin is down-regulated by BCR-ABL, but not by TEL-PDGFRB, however, the result was not validated by Western blotting [
64]. Expression of Scinderin in MEG-01 cells promoted platelet formation and inhibited tumor formation when MEG-01/Scinderin cells were injected into nude mice. MEG-01 is a CML cell line, however, the role of scinderin in BCR-ABL function was not examined in this study [
65]. Nevertheless, our results lead to the hypothesis that suppression of scinderin may represent a mechanism by which BCR-ABL, TEL-PDGFRB and TEL-JAK2 promote leukemic transformation.
ID1, (Inhibitor of DNA binding 1), has been previously shown to be induced in samples from CML patients in chronic phase of the disease suggesting its potential role in leukemia development, and our data further support this hypothesis [
32]. Another study confirmed that Id1 was up-regulated in response to various oncogenic tyrosine kinases, including BCR-ABL, TEL-ABL, TEL-PDGFRB, TEL-JAK2, TEL-TRKC, and FLT3-ITD [
66]. Moreover, inhibition of each of those kinases in leukemic cell lines resulted in significant reductions in Id1 expression, suggesting that Id1 is a common downstream target of deregulated kinases. Id proteins are dominant negative regulators of Helix-loop-helix transcription factors that govern growth and development in mammals [
67]. Previously, it has been shown that Id1 can be regulated by C/EBPβ [
68], and we detected an elevated level of the Cebpb transcript 1 week after BCR-ABL activation. Other immediate early gene(s) may be involved in mediating the observed induction of Id1.
Finally, Stat1 and Gbp1 were uniquely induced by TEL-PDGFRB. Tyrosine phosphorylation of Stat1 and DNA binding has been previously observed downstream of BCR-ABL [
5,
7], TEL-PDGFRB [
11] and TEL-JAK2 [
10]. Interestingly, the induction of Stat1 occurred relatively late (after 24 h) in response to TEL-PDGFRB activation. Given that TEL-PDGFRB also induced a large number of IFN-regulated genes including Gbp1, it is tempting to speculate that there may exist an autocrine loop of IFN-γ, and that this in turn induces the transcription of Stat1 and Gbp1. In support of this hypothesis, Ohmine et al. reported the elevated levels of transcription of IFN-related genes including IFN-γ and IFN-γ receptor two in CML cells from patients in the chronic phase [
34]. Moreover, Advani et al. observed that transcription of a group of IFN-inducible genes was significantly increased in primary mouse bone marrow cells expressing p185 BCR-ABL compared to those expressing p210 BCR-ABL, and this was accompanied by an induction of IFN-γ [
36]. However, neither IFN-γ nor its receptors were induced at any of the time points examined in our experiments. It will be interesting to determine whether Stat1 exerts a direct transcriptional effect in TEL-PDGFRB transformed cells, independent of IFN-γ. In terms of its role during leukemogenesis, however, Stat1 was shown to be dispensable in a myeloproliferative disease induced by Tel-PDGFRB [
69].
In summary, our study revealed that BCR-ABL, TEL-PDGFRB and TEL-JAK2 trigger distinct, yet overlapping changes in gene expression. Several genes that are known to be involved in regulating cell migration, proliferation and differentiation were significantly induced or suppressed at steady state as determined by Q-PCR, giving rise to a hypothesis that their regulation downstream of the three fusions may contribute to leukemogenesis. Further characterization of the genes identified in our study may enhance our understanding of the molecular basis underlying the specificity of diseases caused by the three fusion proteins.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
HK, SY and KH performed experiments, SD and DLB directed the study, HK, LCG, JDJ and DLB wrote the paper. All authors read and approved the final manuscript.