An interstitial deletion on chromosome 4q12 resulting in the formation of the FIP1L1-PDGFRA fusion protein is involved in the pathogenesis of imatinib-sensitive chronic eosinophilic leukemia. The molecular mechanisms underlying the development of disease are largely undefined. Human CD34+ hematopoietic progenitor cells were used to investigate the role of FIP1L1-PDGFRA in modulating lineage development. FIP1L1-PDGFRA induced both proliferation and differentiation of eosinophils, neutrophils, and erythrocytes in the absence of cytokines, which could be inhibited by imatinib. Whereas expression of FIP1L1-PDGFRA in hematopoietic stem cells and common myeloid progenitors induced the formation of multiple myeloid lineages, expression in granulocyte-macrophage progenitors induced only the development of eosinophils, neutrophils, and myeloblasts. Deletion of amino acids 30 to 233 in the FIP1L1 gene [FIP1L1(1–29)-PDGFRA] gave rise to an intermediate phenotype, exhibiting a dramatic reduction in the number of erythrocytes. FIP1L1-PDGFRA and FIP1L1(1–29)-PDGFRA both induced the activation of p38 and extracellular signal-regulated kinase 1/2 (ERK1/2) in myeloid progenitors, whereas signal transducers and activators of transcription 5 (STAT5) and protein kinase B/c-akt were only activated by FIP1L1-PDGFRA. Dominant-negative STAT5 partially inhibited FIP1L1-PDGFRA–induced colony formation, whereas combined inhibition of phosphatidylinositol-3-kinase and ERK1/2 significantly reversed FIP1L1-PDGFRA–induced colony formation. Taken together, these results suggest that expression of FIP1L1-PDFGRA in human hematopoietic progenitors induce a myeloproliferative phenotype via activation of multiple signaling molecules including phosphatidylinositol-3-kinase, ERK1/2, and STAT5. [Cancer Res 2007;67(8):3759–66]
- chronic eosinophilic leukemia
- signal transduction
Idiopathic hypereosinophilic syndromes are a rare heterogenous group of hematologic disorders characterized by an unexplained persistent eosinophilia exceeding 1.5 × 109 eosinophils per liter for more than 6 months in combination with symptoms of organ damage resulting from eosinophil infiltration ( 1). Recent studies, in a subgroup of patients with hypereosinophilic syndrome, identified a specific interstitial 800 kb deletion on chromosome 4q12 [del(4)(q12g12)], resulting in the formation of a fusion protein between a previously uncharacterized gene, FIP1-like 1 (FIP1L1) and the platelet-derived growth factor receptor α (PDGFRA; ref. 2). Sequencing of the fusion gene in several patients revealed that the breakpoint in the PDGFRA gene is conserved and is located in a small region in exon 12, thereby deleting the extracellular and transmembrane domain of the receptor. The breakpoint in FIP1L1, however, is variable and spreads from exon 7 to exon 10 ( 2, 3), resulting in fusion proteins differing more than 120 amino acids in length. The clinical implications of this variability are thus far unknown.
Chromosomal deletion resulting in the formation of FIP1L1-PDGFRA has been observed in 14% to 60% of patients with hypereosinophilic syndrome ( 2– 7). These patients were diagnosed with chronic eosinophilic leukemia (CEL) according to the WHO disease classification criteria and were treated with imatinib, which results in complete remission in many patients with CEL ( 8– 10). Imatinib is a 2-phenylaminopyrimidine derivative designed to inhibit BCR-ABL by association with ATP-binding sites. Imatinib not only inhibits BCR-ABL, but also inhibits the activity of other kinases including c-Kit, c-fms, and PDGFR ( 2, 11– 13). Although it has been shown that FIP1L1-PDGFRA acts as a constitutively active tyrosine kinase resulting in the activation of signal transducers and activators of transcription 5 (STAT5; ref. 2), the molecular mechanisms underlying FIP1L1-PDGFRA–mediated CEL, including the relevance of STAT5 activation, are at the moment, incompletely understood.
Recent studies showed that transplantation of interleukin (IL)-5 transgenic mouse hematopoietic progenitors ectopically expressing FIP1L1-PDGFRA induces a CEL-like phenotype in mice, including tissue eosinophilia, as observed in humans ( 14). However, these studies focused on the effect of expression of the fusion protein on murine hematopoietic progenitors rather than human hematopoietic progenitors, and did not identify the molecular mechanisms underlying the development of disease.
We have therefore investigated whether the expression of FIP1L1-PDGFRA in primary human hematopoietic progenitors was sufficient to induce hypereosinophilia. In addition, we have investigated which signal transduction pathways are aberrantly regulated in primary human hematopoietic progenitors expressing FIP1L1-PDGFRA and whether these molecules are critical for the transforming capacity of this fusion protein. Our data shows that expression of FIP1L1-PDGFRA induces the activation of multiple signaling pathways including STAT5, extracellular signal-regulated kinase (ERK)1/2, p38 mitogen-activated protein kinase (MAPK), and protein kinase B (PKB/c-akt), resulting in cytokine-independent colony formation. Deletion of amino acids 30 to 233 of FIP1L1 prevents the activation of PKB and STAT5, resulting in an intermediate phenotype, suggesting that FIP1L1 may play an important role in the regulation of disease phenotype. These data have implications for the development of novel therapies targeting imatinib-resistant CEL.
Materials and Methods
Isolation and culture of human CD34+ cells. Mononuclear cells were isolated from human umbilical cord blood by density centrifugation over a Ficoll-Paque solution (density, 1.077 g/mL). MACS immunomagnetic cell separation (Miltenyi Biotech, Auburn, CA) using a hapten-conjugated antibody against CD34, which was coupled to beads, was used to isolate CD34+ cells. CD34+ cells were cultured in Iscove's modified Dulbecco's medium (Life Technologies, Paisley, United Kingdom) supplemented with 8% FCS, 50 μmol/L of β-mercaptoethanol, 10 units/mL of penicillin, 10 μg/mL of streptomycin, and 2 mmol/L of glutamine at a density of 0.3 × 106 cells/mL. Cells were cultured either in the presence of stem cell factor (SCF; 50 ng/mL), and FMS-related tyrosine kinase 3 (FLT-3) ligand (50 ng/mL), or in the presence of SCF (50 ng/mL), FLT-3 ligand (50 ng/mL), granulocyte macrophage colony-stimulating factor (GM-CSF; 0.1 nmol/L), IL-3 (0.1 nmol/L), and IL-5 (0.2 nmol/L) to induce eosinophil differentiation. Retroviral transduction experiments were done 2 days after isolation. Cord blood samples were collected from healthy donors after informed consent was provided according to the Declaration of Helsinki. Protocols were approved by the local ethics committee of the University Medical Center in Utrecht.
Viral transduction of CD34+ cells. Bicistronic retroviral DNA constructs were used coexpressing enhanced green fluorescent protein (eGFP) and either FIP1L1-PDGFRA, fusing the NH2-terminal 233 amino acids of FIP1L1 to the COOH-terminal 523 amino acids of PDGFRA, or FIP1L1(1–29)-PDGFRA, a fusion gene in which amino acids 30 to 233 of FIP1L1 were deleted ( 15). Retrovirus was produced by transient transfection of the retroviral packaging cell line, Phoenix-ampho, by calcium phosphate coprecipitation. Cells were plated in 6 cm dishes, 24 h before transfection. A total of 10 μg of DNA was used per transfection. Medium was refreshed 16 h after transfection. After an additional 24 h, viral supernatants were collected and filtered through a 0.2-μm filter. CD34+ cells were transduced in 24-well dishes precoated with 10 μg/cm2 of recombinant human RetroNectin (Takara, Otsu, Japan) for 2 h. Transduction was done with the addition of 0.5 mL of viral supernatant to 0.5 mL of medium containing 0.5 × 106 cells. Twenty-four hours after transduction, 0.7 mL of medium was removed from the cells and 0.5 mL of fresh virus supernatant was added together with 0.5 mL of fresh medium.
Isolation of myeloid progenitors. Hematopoietic progenitors were isolated as described by Manz et al. ( 16). In short, CD34+ cells were isolated as described above and cultured for 1 day in the presence of SCF and FLT-3 ligand. Cells were subsequently washed and resuspended in PBS/5% FCS and incubated for 30 min on ice with a mixture of antibodies (all from Becton Dickinson, Alphen a/d Rijn, the Netherlands). Lineage markers included CD2, CD3,CD4, CD7, CD8, CD14, and CD235a. Myeloid progenitors are negative for these lineage markers. The lineage negative (Lin−), CD34+, and CD38− populations consist of hematopoietic stem cells (HSC). Lin−, CD34+, CD38+, CD123+, and CD45RA− cells are common myeloid progenitors (CMP), whereas Lin−, CD34+, CD38+, CD123+, and CD45RA+ cells are granulocyte-macrophage progenitors (GMP). HSCs, CMPs, and GMPs were sorted using a FACS ARIA (from Becton Dickinson). Isotype antibody staining was used to ensure sorting of the correct population. Sorting of the different progenitor populations was confirmed by culture of the individual populations in the presence of SCF, FLT3L, GM-CSF, IL-3, IL-5, granulocyte colony-stimulating factor, and erythropoietin to allow differentiation of all lineages in a colony-forming assay as described below.
Colony-forming unit assay. Retrovirally transduced cells were sorted from not-transduced cells by flow cytometry and used in colony-forming unit (CFU) assays. CD34+ cells were plated in Iscove's modified Dulbecco's medium (Life Technologies) supplemented with 35.3% FCS (Hyclone, Logan, UT), 44.4% methylcellulose-based medium called Methocult (StemCell Technologies, Vancouver, Canada), 11.1 μmol/L of β-mercaptoethanol, 2.2 units/mL of penicillin, 2.2 μg/mL of streptomycin, and 0.44 mmol/L of glutamine at a density of 1,250 cells/well. CFU assays were done either in the absence of cytokines or in the presence of SCF (50 ng/mL), FLT-3 ligand (50 ng/mL), GM-CSF (0.1 nmol/L), IL-3 (0.1 nmol/L), and IL-5 (0.2 nmol/L). Colonies were scored after 12 days of culture. CFU-GEMM (granulocyte/erythrocyte/monocyte/megakaryocyte), CFU-GM (granulocyte/macrophage), CFU-E (erythrocyte), and CFU-Eo (eosinophil) were scored by May-Grunwald Giemsa staining of cells derived from individual colonies.
Western blot analysis. Western blot analysis was done using standard techniques. In brief, differentiating granulocytes were lysed in Laemmli buffer [0.12 mol/L Tris-HCl (pH 6.8), 4% SDS, 20% glycerol, 0.05 μg/μL bromophenol blue, and 35 mmol/L β-mercaptoethanol], and boiled for 5 min. Equal amounts of total lysate were analyzed by 10% SDS-PAGE. Proteins were transferred to Immobilon-P and incubated with blocking buffer (TBS/Tween 20) containing 5% low-fat milk for 16 h at 4°C before incubating with antibodies against either PDGFRA (Santa Cruz, Inc., Santa Cruz, CA), phosphorylated PKB (Cell Signaling Technology, Danvers, MA), phosphorylated ERK1/2 (Cell Signaling Technology), phosphorylated STAT5a (Cell Signaling Technology), phosphorylated p38MAPK (Cell Signaling Technology), or an antibody against tubulin (Sigma-Aldrich, Zwijndrecht, the Netherlands) overnight in the same buffer. Subsequently, blots were incubated with peroxidase-conjugated secondary antibodies for 1 h. Enhanced chemiluminescence was used as a detection method according to the protocol of the manufacturer (Amersham Pharmacia, Amersham, United Kingdom).
Statistics. A Levene's test for equality of variances was done in all experiments. Subsequently, an independent sample t test was done to compare the differences in colony numbers, between the controls and cells transduced with FIP1L1-PDGFRA or FIP1L1(1–29)-PDGFRA. P ≤ 0.05 was considered significant.
FIP1L1-PDGFRA induces cytokine-independent proliferation and differentiation. To determine whether expression of the FIP1L1-PDGFRA fusion gene is sufficient to recapitulate the development of CEL in human hematopoietic progenitors, bicistronic retroviral DNA constructs were used coexpressing eGFP and either FIP1L1-PDGFRA, or FIP1L1(1–29)-PDGFRA ( 15), in which the amino acids 30 to 233 of FIP1L1 were deleted ( 2). Retrovirus was generated and used to transduce umbilical cord blood–derived CD34+ cells. Transduced cells were plated in CFU assays, and colony formation was analyzed after 12 days. The expression of FIP1L1-PDFGRA in human CD34+ cells dramatically induced colony formation in the absence of cytokines, whereas the addition of IL-3 and IL-5, normally required to induce eosinophil differentiation ( 17), did not further enhance colony numbers ( Fig. 1A ). The expression of both FIP1L1-PDGFRA and FIP1L1(1–29)-PDGFRA induced not only the development of eosinophils but also of erythrocytes and neutrophils both in the absence of cytokines or in the presence of IL-3 and IL-5 ( Fig. 1B–D), indicating that expression of the fusion protein itself induces a myeloproliferative phenotype rather than CEL, this suggests that additional factors are required for the development of the observed clinical phenotype.
Although the expression of the fusion proteins rendered the cells cytokine-independent, the cells remained, at least in part, cytokine-responsive. Stimulation of cells expressing FIP1L1-PDGFRA with IL-3 and IL-5 induced the number of CFU-Eo colonies ( Fig. 1D) without affecting total colony numbers ( Fig. 1A), indicating that IL-5 was still important for the induction of eosinophil development in FIP1L1-PDGFRA–expressing cells.
The role of FIP1L1 in the regulation of FIP1L1-PDGFRA–mediated transformation is, at the moment, incompletely understood. Whereas previous studies show that fusion of the first 29 amino acids of FIP1L1 to the intracellular domain of the PDGFRA gene is both necessary and sufficient to induce cytokine-independent proliferation in Ba/F3 cells ( 2), recent studies suggest that FIP1L1 is not required to induce cytokine-independent proliferation of these cells ( 15). To investigate whether expression of the FIP1L1(1–29)-PDGFRA fusion protein in human CD34+ cells gives rise to a similar phenotype compared with the expression of FIP1L1-PDGFRA, CD34+ cells were retrovirally transduced to ectopically express either FIP1L1(1–29)-PDGFRA or FIP1L1-PDGFRA. Transduced cells were plated in CFU assays, and colony formation was analyzed after 12 days. Although expression of FIP1L1(1–29)-PDGFRA in primary human hematopoietic progenitors also induced colony formation in the absence of cytokines, the number of colonies was reduced compared with cells expressing FIP1L1-PDGFRA ( Fig. 1A), which was predominantly caused by a significant reduction in the number of erythrocyte colonies ( Fig. 1B). No significant difference was observed in both CFU-GM and CFU-Eo numbers ( Fig. 1C and D).
Partial deletion of FIP1L1 resulted in reduced erythrocyte development compared with FIP1L1-PDGFRA, suggesting that FIP1L1 itself is important in the development of the myeloproliferative phenotype observed.
Imatinib inhibits FIP1L1-PDGFRA–induced colony formation, enabling cells to undergo normal differentiation. Patients diagnosed with FIP1L1-PDGFRA–mediated CEL are currently treated with imatinib, often resulting in complete remission. In order to investigate whether imatinib inhibits FIP1L1-PDGFRA–mediated induction of colony formation, CD34+ cells were retrovirally transduced to ectopically express FIP1L1-PDGFRA. Transduced cells were plated into CFU assays either in the absence or presence of imatinib, and colony formation was analyzed after 12 days. Treatment of human hematopoietic progenitors expressing FIP1L1-PDGFRA with imatinib completely blocked colony formation in the absence of cytokines, but not in the presence of IL-3 and IL-5 ( Fig. 2A ). Normal eosinophil development was observed in FIP1L1-PGFRA expressing cells cultured in the presence of IL-3, IL-5, and imatinib compared with control cells cultured under the same conditions ( Fig. 2D). This indicates that imatinib treatment, rather than inducing apoptosis of transduced cells, enables cells to normally respond to cytokines. Surprisingly, the formation of granulocyte-macrophage colonies from CD34+ cells was also inhibited upon treatment with imatinib, suggesting that long-term treatment of patients might also affect normal hematopoiesis.
Expression of FIP1L1-PDGFRA in a specific myeloid progenitor population induces lineage-specific myeloproliferation. The expression of FIP1L1-PDGFRA in human CD34+ hematopoietic progenitors induces a myeloproliferative phenotype rather than a phenotype resembling CEL. The cell surface sialomucin-like adhesion molecule CD34 is expressed on early hematopoietic cells and is generally used as a marker for HSCs ( 18). However, CD34 is not exclusively expressed on HSCs, but is also expressed in more committed cells such as CMP, GMP, and megakaryocyte-erythrocyte progenitors ( 16). To investigate whether the expression of FIP1L1-PDGFRA in different hematopoietic progenitor populations gives rise to different phenotypes, hematopoietic progenitors, including HSCs, CMPs, and GMPs were sorted from CD34+ cells by flow cytometry ( 16), and subsequently retrovirally transduced to ectopically express FIP1L1-PDGFRA. Transduced cells were plated into CFU assays without any cytokines and colony formation was analyzed after 12 days ( Fig. 3A ). Sorting of the different progenitor populations was confirmed by culturing the individual populations in the presence of SCF, FLT3L, GM-CSF, IL-3, IL-5, granulocyte colony-stimulating factor, and erythropoietin to allow differentiation of all lineages in a colony-forming assay (data not shown). The expression of FIP1L1-PDGFRA in HSCs and CMPs resulted in the formation of all myeloid lineages in the absence of cytokines including high numbers of erythrocyte colonies, whereas expression of the fusion protein in a Lin−, CD34+, CD38+, CD45RA−, CD123lo GMP ( 16) population only resulted in high numbers of myeloblasts, neutrophils, and eosinophils ( Fig. 3B). This supports the idea that for FIP1L1-PDGFRA to induce CEL, the translocation is more likely to occur in a more committed population such as the GMP.
FIP1L1-PDGFRA–induced activation of multiple signal transduction pathways. The observed differences in FIP1L1-PDGFRA and FIP1L1(1–29)-PDGFRA–mediated colony formation and lineage development suggests differences in activation of intracellular signal transduction pathways. It has previously been shown that FIP1L1-PDGFRA induces the activation of STAT5 in Ba/F3 cells, whereas ERK1/2 did not seem to be a downstream target of FIP1L1-PDGFRA in this cell line ( 2). To investigate which signal transduction pathways are aberrantly regulated in FIP1L1-PDGFRA expressing human CD34+ cells, and to better understand the molecular mechanisms underlying the observed differences in lineage development caused by the deletion of amino acids 30 to 233 in FIP1L1, CD34+ hematopoietic progenitors were retrovirally transduced to ectopically express FIP1L1-PDGFRA or FIP1L1(1–29)-PDGFRA. Transduced cells were sorted from nontransduced cells and cell lysates were prepared. Expression of FIP1L1-PDFGRA induced phosphorylation of STAT5a, ERK1/2, p38MAPK, and PKB in the absence of cytokines ( Fig. 4 ). Interestingly, whereas expression of FIP1L1(1–29)-PDGFRA resulted in the activation of ERK1/2 and p38MAPK, STAT5a and PKB were not, or were undetectably phosphorylated, providing a potential mechanism for the differences observed in colony formation between both fusion proteins.
STAT5 is necessary but not sufficient for FIP1L1-PDGFRA–induced colony formation. To investigate whether activation of STAT5a in CD34+ cells ectopically expressing FIP1L1-PDGFRA plays an important role in the induction of cytokine-independent colony formation, CD34+ cells expressing FIP1L1-PDGFRA were retrovirally transduced with STAT5aΔ750, a dominant-negative STAT5a ( 19). FIP1L1-PDGFRA–mediated colony formation was inhibited upon the expression of STAT5aΔ750 ( Fig. 5A ), however, dominant-negative STAT5a was insufficient to completely block colony formation, suggesting that other signal transduction pathways may play a critical role. This is supported by the observation that whereas FIP1L1(1–29)-PDGFRA does not activate STAT5, it is still able to induce some degree of myeloproliferation.
Several studies have shown that activation of STAT5a was critical for the induction of differentiation in various hematopoietic lineages including eosinophils, neutrophils ( 20), and erythrocytes ( 21). To investigate whether constitutive activation of STAT5a was indeed sufficient to induce cytokine-independent colony formation from hematopoietic progenitors, CD34+ cells were retrovirally transduced to ectopically express constitutively active STAT5a (STAT5a1*6; ref. 22) or eGFP as a control. Transduced cells were sorted from the nontransduced cells and plated in methylcellulose in the absence of cytokines. Constitutive activation of STAT5a induced the number of colonies ( Fig. 5B), including erythrocyte colonies ( Fig. 5C). Colony size, however, was dramatically reduced compared with FIP1L1-PDGFRA–expressing colonies (data not shown), suggesting that although STAT5a can play a role in FIP1L1-PDGFRA–induced colony formation, the activation of additional signaling molecules is required for the myeloproliferative phenotype observed.
Combined inhibition of phosphatidylinositol-3-kinase and ERK1/2 blocks FIP1L1-PDGFRA–induced colony formation. To investigate whether inhibition of additional signaling pathways could rescue the FIP1L1-PDGFRA–mediated phenotype, CD34+ cells were retrovirally transduced to express FIP1L1-PDGFRA. Transduced cells were plated in methylcellulose medium either in the absence or in the presence of LY294002, SB203580, and U0126, pharmacologic inhibitors of phosphatidylinositol-3-kinase (PI3K), p38MAPK, and ERK1/2, respectively. Whereas colony formation of FIP1L1-PDGFRA–expressing cells was unaffected by the inhibition of p38MAPK, inhibition of either PI3K or ERK1/2 modestly decreased colony numbers both in presence or absence ( Fig. 6B ) of IL-3 and IL-5 ( Fig. 6A). Importantly, combined inhibition of both signaling pathways dramatically reduced colony formation, suggesting that combined activation of PI3K, ERK1/2, and STAT5a is critical for FIP1L1-PDGFRA–mediated cytokine-independent lineage development.
In this study, we have investigated for the first time, the molecular mechanisms underlying FIP1L1-PDGFRA transforming ability in human hematopoietic progenitor cells. Our data show that expression of FIP1L1-PDGFRA in human hematopoietic progenitors induces not only eosinophil differentiation, but also the development of other myeloid lineages including erythrocytes and neutrophils, resembling a myeloproliferative phenotype. Recently, it has been shown that transplantation of mouse hematopoietic progenitors from IL-5 transgenic mice ectopically expressing FIP1L1-PDGFRA induces hypereosinophilia in mice including organ infiltration ( 14). In contrast, transplantation of normal mouse hematopoietic progenitors expressing FIP1L1-PDGFRA induces a myeloproliferative phenotype, suggesting that IL-5 is critical for induction of CEL. In contrast to these observations, although the addition of IL-5 induced eosinophil differentiation of human CD34+ cells expressing FIP1L1-PDGFRA in our experiments, the total number of eosinophil colonies was actually slightly decreased compared with control cells cultured in the presence of IL-5 ( Fig. 1F), suggesting that IL-5 may help induce the phenotype, but is not sufficient.
Besides prolonged hypereosinophilia resulting in end-organ damage, patients with CEL also show increased serum tryptase levels and increased levels of atypical mast cells, distinct from systemic mastocytosis. In addition, distinct features including anemia, thrombocytopenia, neutrophilia, splenomegaly, marrow hypercellularity, and early myeloid precursors in peripheral blood smear have been observed in these patients ( 23, 24). Our experiments show that expression of FIP1L1-PDGFRA in HSCs and CMPs results in the formation of all myeloid lineages, whereas expression of the fusion protein in a Lin−, CD34+, CD38+, CD45RA−, and CD123lo GMP ( 16) population only resulted in high numbers of myeloblasts, neutrophils, and eosinophils without inducing erythrocyte development ( Fig. 3). This supports the idea that the leukemic stem cell for CEL resides in a more committed population such as the GMP. However, it has recently been described that the FIP1L1-PDGFRA fusion gene can be found in both myeloid and lymphoid cells ( 25), indicating that the initial mutation resulting in the expression of FIP1L1-PDGFRA may also occur in HSCs. Our experiments suggest that chromosomal translocation resulting in the expression of FIP1L1-PDGFRA would not be sufficient to induce CEL in the absence of secondary factors or mutations. Interestingly, others have shown that in a mouse transplantation model, serial transplantation of the FIP1L1-PDGFRA–mediated disease was only successful using high cell numbers, suggesting that the expression of FIP1L1-PDGFRA does not induce self-renewal of hematopoietic progenitors ( 14). This suggests that although the initial mutation in patients may occur in HSCs, secondary mutations resulting in increased self-renewal capacity should occur in Lin−, CD34+, CD38+, CD45RA−, and CD123lo GMP cells or even more committed eosinophil progenitors. Interestingly, Krivtsov et al. recently described that expression of the MLL-AF9 fusion protein in a committed GMP progenitor was sufficient to induce leukemia in mice. Although retaining their normal GMP phenotype, the self-renewal capacity of these cells was induced on the expression of MLL-AF9, suggesting that reactivation of self-renewal in committed progenitors is both possible and sufficient to induce leukemia ( 26).
Patients diagnosed with FIP1L1-PDGFRA–mediated CEL are currently treated with imatinib. Although imatinib results in complete remission in many patients ( 8– 10), clinical observations suggest that imatinib also affects normal hematopoiesis. For example, grades 3 to 4 neutropenia has been observed in 14% of Philadelphia chromosome–positive patients treated with imatinib, whereas thrombocytopenia and anemia has been observed in 8% and 3% of these patients, respectively ( 27). Several in vitro studies confirm that imatinib can affect normal hematopoiesis. For example, both differentiation of dendritic cells from mobilized peripheral blood–derived CD34+ cells ( 28), as well as proliferation of T cells ( 29), are inhibited by imatinib. The normal expansion of CD34+ cells is also inhibited upon treatment with imatinib ( 30, 31). Imatinib, in our experiments, does not induce the apoptosis of FIP1L1-PDGFRA–expressing cells, but apparently inactivates the constitutively active tyrosine kinase, enabling cells to undergo normal differentiation ( Fig. 2). This is important for current clinical therapies and indicates that imatinib treatment should be continued on remission to prevent relapse of the disease. However, we also observed that the formation of granulocyte-macrophage colonies from normal CD34+ cells was inhibited on treatment with imatinib ( Fig. 2), suggesting that long-term treatment of patients with CEL using imatinib may ultimately result in the development of additional hematological disorders in these patients.
Expression of FIP1L1(1–29)-PDGFRA in primary human hematopoietic progenitors induced colony formation in the absence of cytokines, however, the number of erythrocyte colonies was reduced ( Fig. 1). No differences were observed in granulocyte-macrophage and eosinophil colony formation. This indicates that FIP1L1 plays a role in cytokine-independent development of some, but not all, myeloid lineages. Previous studies showing that FIP1L1 was dispensable for cytokine-independent proliferation were done in Ba/F3 cells, a mouse pro-B cell line ( 15).
The function of FIP1L1 is incompletely understood. It could be hypothesized that FIP1L1 plays an important role in the regulation of dimerization of the intracellular PDGFRA domain, normally required for activation, resulting in constitutively active fusion protein. Dimerization effects, however, are unlikely because a lack of dimerization would reduce the activation of all signaling pathways. We have shown that both ERK1/2 and p38MAPK are activated upon expression of FIP1L1-PDGFRA and FIP1L1(1–29)-PDGFR, whereas STAT5a and PKB are only activated upon the expression of FIP1L1-PDGFRA but not FIP1L1(1–29)-PDGFRA ( Fig. 4).
Interestingly, a recent study showed that truncation of the first tryptophan residue in the juxtamembrane domain of PDGFRA was sufficient to induce cytokine-independent proliferation of Ba/F3 cells, suggesting that FIP1L1 does not play a role in FIP1L1-PDGFRA–mediated transformation ( 15). This seems to be in contrast with earlier studies showing that deletion of amino acids 6 to 233 of FIP1L1 results in the expression of a protein unable to induce cytokine-independent transformation of Ba/F3 cells ( 2). In addition, whereas the expression of a PDGFRA deletion mutant consisting of one tryptophan residue was sufficient to induce cytokine-independent proliferation, the level of proliferation of these cells was reduced compared with cells expressing FIP1L1-PDGFRA. This suggests that truncation of the juxtamembrane domain is involved in the induction of transformation, but that FIP1L1 may be critical for disease pathogenesis in CEL. Furthermore, although it is evident that the breakpoint in PDGFRA is located in the juxtamembrane region of most patients, deleting one tryptophan residue, one patient has been described with a breakpoint located before the juxtamembrane domain leaving both tryptophan residues intact ( 3), suggesting that truncation of PDFGRA is not always sufficient to induce CEL. Interestingly, a novel t(4;17)(q12;q21) translocation has recently been discovered in a case of juvenile myelomonocytic leukemia resulting in the formation of a FIP1L1-RARα fusion protein ( 32). Because the dimerization capacity of previously described RARα fusion partners seems to be critical for the induction of leukemic transformation, it is likely that FIP1L1 also plays an important role in the development of disease.
We and others have previously shown that STAT5 plays an important role in the regulation of the development of various myeloid lineages ( 20). In this study, we have shown that STAT5 is sufficient to induce the development of granulocyte-macrophage colonies, erythrocyte colonies, and eosinophil colonies in the absence of cytokines ( Fig. 5). Colony size, however, was dramatically reduced compared with FIP1L1-PDFGRA–expressing cells, indicating that other signal transduction pathways play an important role in the regulation of progenitor expansion. Indeed, combined inhibition of the PI3K and ERK1/2 pathway dramatically reduced FIP1L1-PDGRFA–induced colony numbers ( Fig. 6), suggesting that these pathways play an important role in controlling FIP1L1-PDGFRA–mediated progenitor expansion. Our data suggests that although activation of STAT5 may result in cytokine-independent differentiation, PI3K and ERK1/2 are needed for the expansion of transformed colonies. Indeed, both PI3K-PKB and ERK1/2 have been clearly shown to play critical roles in the regulation of proliferation in a variety of cell systems ( 33, 34).
Taken together, our results show that an interstitial deletion on chromosome 4q12, resulting in the expression of FIP1L1-PDGFRA in human hematopoietic progenitors is sufficient to induce cytokine-independent myeloproliferation in human CD34+ progenitors. Investigation of the molecular mechanisms underlying FIP1L1-PDGFRA–mediated lineage development revealed that combined activation of multiple signaling molecules including PI3K, ERK1/2, and STAT5a plays an important role in the induction of cytokine-independent colony formation.
Imatinib inactivates FIP1L1-PDGFRA, enabling cells to undergo normal differentiation, indicating that imatinib treatment should be continued upon remission to prevent relapse of the disease. However, colony formation of control cells was also inhibited by imatinib, suggesting that long-term treatment of patients with CEL using imatinib affects normal hematopoiesis, which may ultimately result in the development of additional hematologic disorders in these patients. It is therefore important to develop alternatives for treatment with imatinib. Because cancer cells become highly dependent on aberrantly regulated intracellular signaling pathways, inhibition of the FIP1L1-PDGFRA–induced signaling pathways might provide future alternative therapies for patients with imatinib-resistant CEL and may reduce the potential side effects of long-term imatinib treatment.
Grant support: Dutch Cancer Society research grant UU2001-2491 and UU2005-3659.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Received November 13, 2006.
- Revision received January 12, 2007.
- Accepted February 2, 2007.
- ©2007 American Association for Cancer Research.