We investigated genetically affected leukemic cells in FIP1L1-PDGFRA+ chronic eosinophilic leukemia (CEL) and in BCR-ABL1+ chronic myeloid leukemia (CML), two myeloproliferative disorders responsive to imatinib. Fluorescence in situ hybridization specific for BCR-ABL1 and for FIP1L1-PDGFRA was combined with cytomorphology or with lineage-restricted monoclonal antibodies and applied in CML and CEL, respectively. In CEL the amount of FIP1L1-PDGFRA+ cells among CD34+ and CD133+ cells, B and T lymphocytes, and megakaryocytes were within normal ranges. Positivity was found in eosinophils, granulo-monocytes and varying percentages of erythrocytes. In vitro assays with imatinib showed reduced survival of peripheral blood mononuclear cells but no reduction in colony-forming unit growth medium (CFU-GM) growth. In CML the BCR-ABL1 fusion gene was detected in CD34+/CD133+ cells, granulo-monocytes, eosinophils, erythrocytes, megakaryocytes and B-lymphocytes. Growth of both peripheral blood mononuclear cells and CFU-GM was inhibited by imatinib. This study provided evidence for marked differences in the leukemic masses which are targeted by imatinib in CEL or CML, as harboring FIP1L1-PDGFRA or BCR-ABL1.
BCR-ABL1+ chronic myeloid leukemia (CML) is a paradigm for the multilineage involvement of an imatinib-sensitive chronic myeloproliferative disorder in which an affected stem cell ordinately differentiates toward both myeloid and lymphoid hematopoietic pathways.
Chronic eosinophilic leukemia (CEL), is a chronic myeloproliferative disorder whose diagnosis is based on persistent eosinophilia (>1.5 × 109/l), organ involvement and increased blast cells in peripheral blood or bone marrow and/or clonality, as shown by cytogenetics and X-inactivation.1, 2, 3, 4 Cools et al.5 first described a genetic clonal hallmark for a specific CEL subgroup. The CHIC2 gene deletion at 4q12 results in a fusion between the FIP1L1 and PDGFRA genes. As a consequence the PDGFRA tyrosine kinase is activated. Interestingly, imatinib mesylate, a tyrosine kinase inhibitor, rapidly induces remission in the majority of patients, although disease is not fully eradicated.
In order to compare the hematopoietic cell lineages affected by FIP1L1-PDGFRA in CEL and by BCR-ABL1 in CML, we combined fluorescence in situ hybridisation (FISH) with cytomorphology (MISH), as well as with immunostaining fluorescence immunophenotyping and interphase cytogentics as a tool for the investigation of neoplasms (FICTION). Furthermore, we tested in vitro imatinib sensitivity both on peripheral blood mononuclear cells and in colony-forming unit growth medium (CFU-GM) from patients with CEL and CML. Results showed that FIP1L1-PDGFRA+ cells are limited to advanced stages of myeloid cell differentiation, whereas BCR-ABL1 is present in both myeloid and lymphoid lineages. Accordingly, imatinib inhibited peripheral blood mononuclear cells and CFU-GM in BCR-ABL1 +CML, but only peripheral blood mononuclear cells in FIP1L1-PDGFRA+ CEL. In this study, we assessed a restricted lineage affiliation of hematopoietic cells labeled by FIP1L1-PDGFRA in CEL.
Materials and methods
Patients and controls
A total of 16 FIP1L1-PDGFRA-positive CEL cases were included in this study. Bone marrow aspirates were used to investigate lineage affiliation in eight patients. In eight additional cases, peripheral blood was used for in vitro studies on imatinib sensitivity. Patients were selected from the Hematology Departments of the Universities of Perugia and Bari (Italy), from the Hematology Department and Centre for Human Genetics of the University of Leuven (Belgium) and from the Wessex Regional Genetics Laboratory of the Salisbury District Hospital, Salisbury (UK). Four cases of CML were selected from patients referred to the Hematology Department of the University of Perugia (Italy). Twelve healthy donors were used as controls. The study was approved by the Hematology Department IRB (Reg. no. 00003450; FWA 00005268) of University of Perugia, Italy.
Genomic probes and monoclonal antibodies
The FIP1L1-PDGFRA/4q12 was studied with clone RP11-3H20 labeled with biotin alone or in combination with clone RP11-120K16 labeled with digoxigenin (RP11 belongs to the Roswell Park Cancer Institute (RPCI)11 library, http://bacpac.chori.org/). The t(9;22)(q34;q11)/BCR-ABL1 was investigated with the LSI BCR-ABL dual-color/dual-fusion translocation probe (BCR clone in green and ABL1 clone in orange) (Vysis, Olympus, Milan, Italy).The following monoclonal antibodies were used: anti-CD13, anti-CD14, anti-CD33, anti-CD3, anti-CD7, anti-CD19, anti-CD20, anti-glycophorin C, anti-glycophorin A (Dako, Milan, Italy), anti-CD34 (Becton-Dickinson, Milan, Italy), anti-CD133 (Miltenyi Biotec S.r.l., Bologna, Italy).
FICTION and MISH
A FICTION published methodology6 was slightly modified. Briefly, cytospins were prepared from bone marrow mononuclear cells after centrifugation on Lymphoprep (AXIS-SHIELD, Norway) using 100 μl from a cell suspension of 1 × 106/ml. Slides were air-dried at room temperature for 24 h fixed in acetone for 10 min and incubated with monoclonal antibodies for 30 min at room temperature. The three-step staining technique used the following Cy3-conjugated polyclonal antibodies: goat anti-mouse, rabbit anti-goat and donkey anti-rabbit (Jackson Immunoresearch/Li StarFISH, Milano, Italy). After immunostaining, slides were fixed in Carnoy's fixative (metanol:acetic acid, 3:1) for 1 min and in 1% paraformaldehyde for 10 min, washed in distilled water and dehydrated in an ethanol series. For MISH investigations, bone marrow smears were air-dried and fixed in methanol:acetic acid (3:1) for 5 min. Slides and probes were co-denaturated on a hot plate at 76°C and incubated overnight at 37°C in a humidified chamber. To detect biotinylated DNA, slides were incubated three times: with fluorescein isothiocyanate (FITC)-conjugated avidin, with biotinylated goat anti-avidin antibody and with FITC-conjugated avidin (Vector Laboratories, DBA Italia, Milan, Italy).
Immunophenotype and hybridization signals were simultaneously identified and counted under an Olympus fluorescence microscope with filter sets for Cy-3 and FITC equipped with a CCD camera (Sensys-Photometrics, Tucson, AZ, USA) run from image analysis software (Vysis, SmartCapture, Olympus, Milan, Italy). For each antibody, at least 13 (range 13–94) cells were checked in CEL and CML patients and at least 145 cells in normal controls (range 145–375).
MISH on eosinophils, and on megakaryocytes was carried out after morphological identification of single cells on bone marrow smears applying a set of DNA clones described previously (Figures 1 and 2). All the experiments were evaluated by two independent observers.
CD34+ cell selection
CD34+ stem cells were labeled with CliniMACS CD34 microbeads (Miltenyi Biotec S.r.l., Bologna, Italy) and passed through the magnetic field of a Mini MACS separator (CD 34 Progenitor Cell Isolation Kit, Miltenyi Biotec S.r.l., Bologna, Italy). Purity was analyzed on a FACscan (Cytomics FC 500, Beckman Coulter, Milan, Italy) using a monoclonal antibody specific for CD34 conjugated with FITC (Becton Dickinson, Milan, Italy). After suspension in phosphate-buffered saline, cell cytospins were prepared and fixed: (1) 5 min in 30% fixative (3:1 methanol:acetic acid) diluted with 0.075 M KCl, (2) 10 min in 20% ethanol diluted with 0.075 M KCl and (3) 10 min in fixative (3:1 methanol:acetic acid). After fixation interphase FISH was performed using clone RP11-3H20 and clone RP11-120K16, as described previously.7
Progenitor cell colony assay
Thawed bone marrow mononuclear cells were washed in Iscove-modified Dulbecco's medium (IMDM; StemCell Tecnologies, Vancouver, Canada). Cells were counted and 1 × 105 were plated in 1 ml aliquots in 30 mm Petri dishes in a semisolid assay. At least six Petri dishes were prepared for each patient. The medium contained 30% FCS (HyClone, Logan, UT, USA), 3 U/ml rhuEpo, 50 ng/ml stem cell factor, 10 ng/ml GM-CSF, 10 ng/ml IL-3 (all from PeproTech, UK), 0.9% methylcellulose (StemCell Inc., Canada) and IMDM. Cultures were incubated for 14 days, at 37°C, 5% CO2, in a humidified incubator. CFU-GM colonies were scored under an inverted microscope according to standard criteria.8 Colonies (40–50) were transferred singly into a microtiter well containing 40 μl hypotonic solution (0.075 M KCL) for each cytospin. Cytospins were fixed following the same protocol described before for FISH on CD34+ cells. After fixation interphase FISH was carried out applying a set of DNA clones described previously in the genomic probe section. All the experiments were evaluated by two independent observers.
Clonality by X-chromosome inactivation
Both peripheral blood and buccal epithelial samples were obtained from the single CEL woman included in the study. Mononuclear cells were isolated by Ficoll–Hypaque density gradient centrifugation (Lympholyte-H, CEDARLANE, HORNBY, Canada). T lymphocytes were purified from the mononuclear cell fraction using magnetic immunobeads coated with an anti-CD3 antibody (Miltenyi Biotech GmbH, Bergish Gladbach, Germany). Polymorphonuclear leukocytes (PMN) were isolated from the red cell pellet following erythrocytes removal by hypotonic lysis.
The X-chromosome inactivation patterns were established by polymerase chain reaction (PCR) analysis of DNA methylation at the human androgen receptor locus (HUMARA) as previously published.9
In vitro imatinib sensitivity
Test 1: Liquid culture assay
Peripheral blood mononuclear cells or granulocytes from CEL, CML and normal controls were cultured in 24-well plates in 1 ml aliquots of 2–5 × 106 cells in Roswell Park Memorial Institute medium 1640 medium supplemented with 10% serum. Duplicate wells were treated with 1 and 5 μM imatinib. Cells were counted with a hemocytometer twice weekly for 3 weeks. The medium was supplemented on days 7 and 14. A decrease in cell numbers in imatinib-treated wells compared with untreated control wells indicated a positive response.
Test 2: CFU-GM assay
Peripheral blood mononuclear cells from CEL, CML and normal control were separated using lymphoprep (Axis-Shield, Oslo, Norway) and cultured in methylcellulose supplemented with growth factors (Stem Cell Technologies Ltd, Vancouver, Canada) at a cell density of 2 × 105 cells/ml in 3 cm Petri dishes. Imatinib (Novartis, Basel, Switzerland) was added to final concentrations of 1 and 5 μM. Colony numbers were scored on days 7 and 14 from triplicate plates. The response index was calculated as the mean reduction in colony numbers in 1 and 5 μM treated dishes on days 7 and 14 compared with untreated dishes. After control experiments using normal individuals and patients with BCR-ABL1-positive CML, an index below 0.2 was established as indicating positive response.
The outcomes of FICTION and MISH studies on CEL patients were summarized in Table 1 (see also Figure 1a and b). CD34 was investigated in 5/8 cases (patients 1, 2, 4, 5, 6, Table 1). In all of them the percentage of CHIC2 deletion was equal to or below the cutoff (13%) established at the upper limit on normal control (Table 1). In patient 8, purified CD34+ cells were investigated with microbeads conjugated to an anti-CD34 monoclonal antibody. FISH analysis showed one signal in 15/400 cells (4%) of the enriched CD34+ population (normal experiment control 5%, i.e., 18/400). We used the anti-CD133 monoclonal antibody as an additional marker for a totipotent stem cell. This antibody was studied in patients 2 and 6 and the percentage of deletion was 9 and 7%, respectively (Table 1). For this antibody, the cutoff was set at 10%.
CD33 was investigated in 6/8 patients. The percentage of cells bearing CHIC2 deletion ranged from 84 to 95%. CD13 was evaluated in 7/8 patients showing a percentage of deletion between 80 and 98%. The anti-CD14 antibody also showed that 88–97% cells were affected by the CHIC2 deletion in the five patients studied (Table 1).
Glycophorin C was investigated in 5/8 (no. 1, 2, 3, 4, 6, Table 1) patients in whom CHIC2 deletion was detected in 37–88% of cells (normal cutoff 11%). As glycophorin C may crossreact with some granulocytes, glycophorin A was also tested. Positivity confirmed that erythroid cells were affected by CHIC2 deletion. However, CHIC2 deletion in glycophorin A-positive cells was 48% (38/79), 46% (43/93) and 15% (14/94), respectively in patient 1, 2 and 6, suggesting inter-individual variance in the malignant erythrocyte component of CEL.
Interphase fluorescence in situ hybridization (I-FISH) on megakaryocytes from healthy controls showed colocalization of green and red signals as expected in the absence of 4q12 deletion. Results in 30 megakaryocytes in patient 2 showed between 6 and 26 copies of colocalized red–green signals (Table 1 and Figure 1a).
Eosinophils were identified by autofluorescence granules. In all patients, the positivity ranged between 88 and 98% of elements (Table 1).
In four cases of BCR-ABL1+ CML, the BCR-ABL1 rearrangement was observed on totipotent CD34+ and CD133+ stem cells and in committed lineages downstream, that is, granulo-monocytes, erythrocytes, megakaryocytes and B lymphoid cells (Table 2 and Figure 2a and 2b). In patient 2, 80% of eosinophils were positive for BCR-ABL1 rearrangement (Table 2).
FISH on clonogenic cells, that is, CFU-GM colonies, recovered at day 14 in methylcellulose, and was carried out in CEL (no. 1, 2, 6, Table 1) and in CML (no. 2 and 3 Table 2). Results showed no CHIC2 deletion in CEL cases (the percentage of CHIC2 deletion was equal to or above the cutoff of 5%) and positivity for BCR-ABL1 in CML cases (respectively 100% in one case, and 50% in the other).
Clonal analysis in the CEL female patient (case no. 2, Table 1) showed skewed X-chromosome inactivation patterns in both granulocytes and T lymphocytes. To distinguish between the monoclonal nature of these cells and skewed Lyonization, we examined the clonality pattern of buccal mucosa cells as non-hematological control for the myeloid stem cell disorder. The band pattern was opposite to that of PMN and T lymphocyte in keeping with a true clonal hematopoiesis affecting both granulocytes and T cells.
In vitro tests for sensitivity to imatinib in CEL showed that peripheral blood mononuclear cells in liquid cultures were inhibited (Figure 3, bottom left), but no significant response was seen in CFU-GM (median response index: 0.33 vs 0.43 in controls) (Figure 3, bottom right). By contrast, three newly diagnosed BCR-ABL1+ CML patients showed a significant response in both liquid cultures (Figure 3, left) and CFU-GM assay (median response index=0.05) (Figure 3, right).
This study provided us with new insights on the characterization of leukemic populations from two imatinib-sensitive myeloproliferative disorders, namely BCR-ABL1+ CML and FIP1L1-PDGFRA+ CEL.
As expected, in CML the BCR-ABL1 rearrangement consistently labeled multipotential stem cells expressing CD34 and CD133 antigens. Moreover, both myeloid and lymphoid cells were affected, mirroring a predicted cascade of hematopoietic differentiation.10 These findings concur with results of conventional cytogenetics, G6PDH enzymatic activity, X-inactivation and detection of the BCR-ABL1 transcript11, 12, 13, 14 and validate our FISH approach for the assignment of a typical leukemic genetic lesion to different hematopoietic lineages.
From this study, in CEL the CD34+ stem cells were not affected, although in one case (patient 4, Table 1) with a high value of CHIC2 deletion we cannot exclude the presence of FIP1L1-PDGFRA in some CD34+ cells. Moreover, as a low amount of positive nuclei from CD34+ sorted cells were reported by Robyn et al.,15 we also checked FIP1L1-PDGFRA in CD34+ enriched cells after microbeads in patient 8 confirming our FICTION results. Furthermore, in two additional cases (patients 2 and 6, Table 1), CHIC2 deletion was also absent in CD133-positive cells. Thus, in contrast with BCR-ABL1 in CML, CD34+ and CD133+ stem cells are virtually spared by the FIP1L1-PDGFRA genetic lesion in CEL.
B and T lymphocytes were not affected by FIP1L1-PDGFRA, as shown by identical results in CEL and normal samples (Table 1 and Figure 1). However, in our female CEL patient in whom X-inactivation could be performed (patient 2, Table 1), we found that clonality included T lymphocytes, despite the absence of FIP1L1-PDGFRA fusion, suggesting that, at least in some cases, FIP1L1-PDGFRA recombination occurs after a still unknown clonal mutation in a multipotential stem cell. Similar results have been recently shown in a subset of chronic myeloproliferative disorders in which clonality included cells with and without the typical JAK2V617F mutation.16 Interestingly, as far as it concerns lymphocyte cells, analogies are emerging between FIP1L1-PDGFRA in CEL and PDGFRB recombinations in other Ph-negative chronic myeloproliferations. In a previous publication on a case of atypical CML, we showed that the H4-PDGFRB fusion affected myeloid progenitors but neither CD3+ nor CD19+ lymphocytes.17 More recently we obtained similar results in a case of ETV6-PDGFRB+ chronic myelomonocytic leukemia in which both CD3+ and CD19+ cells were normal (Crescenzi B and Mecucci C, unpublished). Whether PDGFRA and PDGFRB recombinations affect a myeloid committed progenitor, or if an affected multipotential stem cell undergoes only restricted and disparate myeloid maturation, when targeted by PDGFRA or PDGFRB recombinations, remains to be determined. With respect to PDGFRB, the last hypothesis is supported by Shigematsu et al. (Blood 2004; 104, abstract no. 387), who reported that an ETV6-PDGFRB fusion gene was able to address only myeloid lineage differentiation when transfected in lymphoid progenitors.
In this study we found that, in addition to eosinophils, the leukemic mass in FIP1L1-PDGFRA+ CEL was limited to erythrocytes and granulomonocytes. Strikingly, CFU-GM growth was inhibited by imatinib in CML, but not in CEL (Figure 3) and, accordingly, FISH on CFU-GM from CML was positive for BCR-ABL1, whereas FISH on CFU-GM from three CEL cases was negative for FIP1L1-PDGFRA. Whether FIP1L1-PDGFRA behaves as a genetic lesion able to impart self-renewal properties to mature precursors remains to be investigated. Alternatively, the restricted FIP1L1-PDGFRA+ hematopoiesis in CEL might be supported by microenvironmental conditions, such as overexpression of specific cytokines. Interestingly, a critical role of extrinsic factors in the pathogenesis of a full CEL phenotype has been recently emphasized in a murine model published by Yamada et al.18 who showed that FIP1L1-PDGFRA was not able to generate organ infiltration by eosinophils in the absence of overexpression of interleukin 5.
In conclusion, for the first time we identified major differences between BCR-ABL1+ cells in CML and FIP1L1-PDGFRA+ cells in CEL. Altogether our results show that the BCR-ABL1+ model of affected CD34+ and CD133+ stem cell ordinately undergoing multilineage differentiation does not apply to CEL whose leukemic mass, labeled by FIP1L1-PDGFRA fusion and targeted by imatinib is restricted to advanced stages of myeloid hematopoiesis.
We thank Dr F Falzetti and Dr T Zei for selection of CD34+ cells; Dr G Guglielmini and Dr A Santucci for preparation of histograms, and Dr GA Boyd for assistance in the preparation of this paper. This study was supported by grants from AIRC (Associazione Italiana Ricerca sul Cancro), Associazione ‘Sergio Luciani’; Fabriano, Italy, Fondazione Cassa di Risparmio, Perugia, Italy, MIUR (Ministero per l’Istruzione, l’Università e la Ricerca Scientifica), the Belgian Programme of Interuniversity Poles of Attraction initiated by Belgian State, Prime Minister's Office, Science Policy Programming and the Leukemia Reasearch Fund, UK. PV is a Senior Clinical Investigator of FWO-Vlaanderen. BC is supported by FIRC (Federazione Italiana Ricerca sul Cancro).