Discovery of a fusion kinase in EOL-1 cells and idiopathic hypereosinophilic syndrome

Abstract

Idiopathic hypereosinophilic syndrome (HES) is a myeloproliferative disease of unknown etiology. Recently, it has been reported that imatinib mesylate (Gleevec), an inhibitor of Bcr-Abl kinase useful in the treatment of chronic myeloid leukemia, is also effective in treating HES; however, the molecular target of imatinib in HES is unknown. This report identifies a genetic rearrangement in the eosinophilic cell line EOL-1 that results in the expression of a fusion protein comprising an N-terminal region encoded by a gene of unknown function with the GenBank accession number NM_030917 and a C-terminal region derived from the intracellular domain of the platelet-derived growth factor receptor α (PDGFRα). The fusion gene was also detected in blood cells from two patients with HES. We propose naming NM_030917 Rhe for Rearranged in hypereosinophilia. Rhe-PDGFRα fusions result from an apparent interstitial deletion that links Rhe to exon 12 of PDGFRα on chromosome 4q12. The fusion kinase Rhe-PDGFRα is constitutively phosphorylated and supports IL-3-independent growth when expressed in BaF3 cells. Proliferation and viability of EOL-1 and BaF3 cells expressing Rhe-PDGFRα are ablated by the PDGFRα inhibitors imatinib, vatalanib, and THRX-165724.

Materials and Methods

Compounds. Imatinib mesylate was extracted from capsules of Gleevec. Vatalanib was prepared according to the published procedure (11). THRX-165724 was prepared by coupling piperazine to the carboxyl group of SU6668 (12).

cDNA Cloning, Plasmids, and Oligos. Total RNA was isolated from 5 × 10⁷ EOL-1 cells by using the RNeasy kit (Qiagen, Valencia, CA). Then 100 ng of total RNA was used to make cDNA in a volume of 20 μl with a reverse oligonucleotide that primes in the 3′ untranslated region of PDGFRα [5′-tccgcattgcaataaagtgg-3′ (bases 3478–3459, GenBank accession no. M22734)] and the Thermoscript RT-PCR system (GIBCO/BRL). Two microliters of the cDNA solution served as template in a 100-μl PCR reaction to amplify full-length Rhe-PDGFRα [forward oligo, 5′-gttgcgctcggggcggccat-3′ (bases 150–169, accession no. NM_030917); reverse oligo, 5′-ttctgaacgggatccagagg-3′ (bases 3456–3437, accession no. M22734)]. The PCR fragment was isolated by agarose gel electrophoresis, cloned into the TOPO PCR vector (Invitrogen), and sequenced. The error-free sequence of a splice variant lacking the two observed alternatively spliced exons was cloned into the mammalian expression vector pcDNA 3.1(+) (Invitrogen). Patient cDNA was generated as described above, using random hexamers or the specific PDGFRα primers 5′-ggatgtcggaatatttagaa-3′ and 5′-gcagaaaggtactgcctttc-3′. To analyze patient cDNA for Rhe-PDGFRα fusion transcripts the following primer pair was used: Rhe forward, 5′-aattatgggtttaatgaag-3′ (bases 651–699, accession no. NM_030917); and PDGFRα reverse, 5′-aactttcatgacaggttgg-3′ (bases 2000–1982, accession no. M22734). For the PCR analysis of the genomic fusion point in EOL-1 as well as the two patients, an oligonucleotide priming 3′ of PDGFRα exon 12 in the reverse orientation was combined with the following specific forward primers: PDGFRα genomic reverse, 5′-ttcttactaagcacaagctcagatc-3′ (bases 13912–13888, accession no. AC098587); EOL-1 and patient 3 genomic forward, 5′-aagcatctaattaggtgaaactg-3′ (bases 48554–48576, accession no. NT_022853); and patient 1 genomic forward, 5′-cagggaagaactggaaactc-3′ (bases 22466–22485, accession no. NT_022853).

Cells and Cell Lines. The EOL-1 and BaF3 cell lines were obtained from the Deutsche Sammlung von Mikroorganismen und Zellkulturen (Braunschweig, Germany). The basic culture medium for the EOL-1 and BaF3 cell lines was RPMI 1640 (GIBCO/BRL) supplemented with 10% FBS, 100 units/ml penicillin, and 100 units/ml streptomycin. The medium for BaF3 cells was also supplemented with 1 ng/ml IL-3 (BioSource International, Camarillo, CA). A BaF3 cell line expressing Rhe-PDGFRα was created by electroporation of BaF3 cells at 300 mV/960 μF. After electroporation, the BaF3 cells were maintained in IL-3-containing medium for 48 h, selected in IL-3-containing medium plus 1 mg/ml G418 for 10 days, and subcloned by limiting dilution.

Cell Viability Assays. Cell viability was assessed by tetrazolium salt reduction using the MTT [3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl tetrazolium bromide] assay (Roche). In a 96-well plate, 5 × 10⁴ cells per well were plated in the presence of serial dilutions of compounds. The cells were incubated for 72 h before the addition of MTT substrate.

Immunoprecipitation and Western Blotting. Antibodies against PDGFRα/β and against phosphotyrosine (4G10) were purchased from Upstate Biotechnology (Lake Placid, NY). For each immunoprecipitation, 1 × 10⁷ cells were lysed in 0.75 ml of modified RIPA buffer [50 mM Tris·HCl, pH 7.4/1% Nonidet P-40/150 mM NaCl/1 mM EDTA/1 mM Na₃VO₄ and protease inhibitor mixture (Roche)]. The lysates were incubated with the appropriate antibody and protein G beads (Sigma) overnight at 4°C. The immunocomplexes were recovered by centrifugation, washed with RIPA buffer, boiled in sample buffer, and resolved by SDS/PAGE. The proteins were transferred to a poly(vinylidene difluoride) (PVDF) membrane (Invitrogen), blocked with PBS/0.1% Tween 20/3% BSA, and probed with a specific antibody for 3 h at room temperature. Subsequently, the blots were washed with PBS/0.1% Tween 20. Specific antibody binding was detected with a horseradish peroxidase-coupled secondary antibody, followed by enhanced chemiluminescence (ECL, Amersham Biosciences) and exposure to film. The primary antibody was typically stripped with ImmunoPure IgG Elution Buffer (Pierce) for reprobing of the blot with a second antibody.

Phosphorylation Inhibition Assay. Cells (1 × 10⁷) were incubated in 3 ml of media with the indicated concentration of drug for 1 h. The cells were subsequently lysed and immunoprecipitated with the appropriate antibody. Then, SDS/PAGE was performed, followed by immunoblotting with the antiphosphotyrosine antibody 4G10.

Protein Digestion and Peptide Analysis. This work was performed by Proteomic Research Services (Ann Arbor, MI). The sample was provided to Proteomic Research Services in the form of 50 μl of protein G immunoaffinity resin, to which was bound tyrosine phosphoproteins from 1 × 10⁸ EOL-1 cells via antibody 4G10. The proteins were fractionated by SDS/PAGE and visualized by staining with SYPRO Ruby (Molecular Probes). Plugs were chosen for excision based on an overlay of the SYPRO-stained lane with that of a companion lane visualized by Western blotting with 4G10. The plugs were subjected to in-gel digestion with trypsin (ProGest, Ann Arbor, MI), and a portion of the supernatant was used for analysis by matrix-assisted laser desorption ionization (MALDI)/MS. MALDI/MS data were acquired on an Applied Biosystems Voyager DE-STR instrument, and the observed m/z values were submitted to a search for peptide mass fingerprints by the software package PROFOUND (Proteometrics, Ann Arbor, MI), querying the NCBInr database. In cases where MALDI/MS analysis was inconclusive, samples were analyzed by nano-liquid chromatography followed by two-dimensional MS (LC/MS/MS) on a Micromass (Manchester, U.K.) Q-Tof2 instrument. The MS/MS data were searched by using the search engine MASCOT from Matrix Science (www.matrixscience.com).

Results

Model Cell Line EOL-1. EOL-1 is a cell line derived from the blood of a 33-year old man with acute eosinophilic leukemia following hypereosinophilic syndrome. The EOL-1 cell line has been characterized as having a hyperdiploid karyotype with a range of 48–51 chromosomes per cell: 50,XY,+4,+6,+8,del(9)(q22),+19. Imatinib and two other inhibitors of the PDGFR tyrosine kinase family (vatalanib and THRX-165724, Fig. 1) potently reduce the viability of EOL-1 cells (Fig. 2). Common targets inhibited by imatinib, vatalanib, and THRX-165724 include PDGFRα and -β, as well as c-Kit (13, 14).

Identification of a Tyrosine Phosphoprotein in EOL-1 Cells. A common pathway for cellular transformation has been shown to result from mutations in protein kinases that lead to their constitutive activation and autophosphorylation. To identify a potentially autophosphorylated mutant kinase in EOL-1 cells, tyrosine phosphorylated proteins were immunoprecipitated from a cell extract and visualized by immunoblotting. This revealed a very prominent phosphoprotein of apparent molecular weight 110 kDa (data not shown). When cells were treated with imatinib before lysis, the prominent 110-kDa phosphoprotein and minor phosphoproteins were no longer observed. The imatinib targets PDGFRα (170 kDa) and -β (190 kDa), as well as c-Kit (145 kDa), have molecular weights that are larger than 110 kDa. To identify the 110-kDa protein, the protein was subjected to in-gel tr yptic digestion. Tr yptic peptides were analyzed by MALDI/MS and LC/MS/MS, and queries for peptide mass fingerprints were made against the NCBInr database. This provided matches to peptides from three proteins: (i) nucleolin, (ii) the protein predicted for the gene with accession number NM_030917, and (iii) PDGFRα. Nucleolin is a 105-kDa phosphoprotein involved in ribosome assembly. It is not a kinase and, therefore, is an unlikely target for imatinib. NM_030917 codes for a protein with unknown function. Although PDGFRα is a known target of imatinib, its apparent molecular weight is 170 kDa rather than 110 kDa. Interestingly, all 10 of the PDGFRα peptides mapped to the C-terminal cytoplasmic region of the receptor that contains the kinase domain, and all 8 peptides of the protein encoded by NM_030917 mapped to its N-terminal region.

Identification of a Genomic Deletion, Fusion Gene, and Fusion Protein in EOL-1 Cells. The genes for NM_030917 and PDGFRα are located in close proximity on chromosome 4q11–12 (separated by ≈800 kb). This proximity raised the possibility that a fusion kinase comprising these two proteins, caused by a submicroscopic rearrangement of a small portion of chromosome 4, exists in EOL-1 cells. To determine whether NM_030917 and PDGFRα are fused in EOL-1, oligonucleotides were designed to prime from the 5′ end of NM_030917 and the 3′ end of PDGFRα. RT-PCR was performed by using total RNA isolated from EOL-1 cells, resulting in the selective amplification of a 2.5-kb product. This fragment was cloned and multiple clones were sequenced. The 2.5-kb product indeed encompasses a NM_030917-PDGFRα fusion transcript (Fig. 3A). Four different splice variants were observed among the sequenced clones, all of which occur in the NM_030917 segment. The fusion gene encodes predicted proteins of 811–849 aa (Fig. 3B). A tryptic peptide (ANENSNIQLPYDSR) spanning the predicted fusion junction was observed by ion spray MS. We have termed the gene represented by NM_030917 “Rhe,” for Rearranged in hypereosinophilia.

By using oligonucleotides that primed 5′ and 3′ of the fused exonic sequences of Rhe and PDGFRα, respectively, a fragment was amplified from genomic DNA that contained the EOL-1 chromosomal breakpoint. Rhe is contained in the genomic sequence NT_022853 from position 1583318–1666068. Seventeen exons have been assigned in this sequence by the NCBI. According to these assignments, the break in EOL-1 occurs in the intron following exon 11 of Rhe and is fused within exon 12 of PDGFRα (15). The fusion transcript is in-frame (Fig. 4A). The splicing process uses the first AG in the PDGFRα exon 12 fragment as a splice acceptor site for intron 11 of Rhe (Fig. 4B).

Identification of Rhe-PDGFRα in HES Patient Cells. To determine whether the Rhe-PDGFRα fusion was present in HES patients, blood cells from four patients diagnosed with HES were obtained (for patient descriptions, see Supporting Text, which is published as supporting information on the PNAS web site, www.pnas.org). Patients 1 and 2 have been treated with imatinib. Patient 1 responded to treatment, but patient 2 did not. After showing a complete hematologic remission, patient 1 relapsed and died. This patient had multiple clonal cytogenetic abnormalities that led to the diagnosis of CEL. Genomic DNA as well as total RNA and cDNA were prepared from all patient cells except for patient 1, from whose cells only genomic DNA (no RNA or cDNA) was obtained before imatinib treatment. The cDNA samples were subjected to PCR with a primer pair spanning the fusion point determined in EOL-1 cells. In the samples from patients 1 (relapsed) and 3, fragments could be amplified from the cDNA that constituted in-frame fusion transcripts between Rhe and PDGFRα (Fig. 4A). No Rhe-PDGFRα fusion was detected in patients 2 and 4. In patient 1, the fusion transcript connects exon 8 of Rhe within exon 12 of PDGFRα. A similar approach as for EOL-1 was used to identify the genomic breakpoint. In patient 1, the intronic break is at an AG dinucleotide that serves as the splice acceptor site so that exon 8 in Rhe and part of exon 12 in PDGFRα are fused in-frame in the fusion transcript (Fig. 4B). The Rhe-PDGFRα fusion in patient 1 was detected in genomic DNA preparations derived from cells taken before the start of imatinib therapy and at the time of relapse. The analysis of Rhe-PDGFRα cDNA at the time of relapse revealed a point mutation in the PDGFRα kinase domain. The mutation affects amino acid position 674 in PDGFRα, resulting in the substitution of threonine by isoleucine (T674I; Fig. 4C).

In patient 3, the fusion transcript and the genomic break are identical to the mutation found in EOL-1 cells (Fig. 4 A and B).

Inhibition of Rhe-PDGFRα Rabbit antiserum raised against a peptide derived from the cytoplasmic domain of PDGFRα successfully immunoprecipitated the 110-kDa phosphoprotein from an EOL-1 cell extract. Treatment of the cells with imatinib, vatalanib, and THRX-165724 resulted in dose-dependent inhibition of phosphorylation of the fusion protein (Fig. 5). In the case of vatalanib and THRX-165724, the potency of inhibition of phosphorylation (IC₅₀ of 300 and 30 nM, respectively) corresponds closely to the potencies in inducing apoptosis (IC₅₀ of 140 and 25 nM, respectively). Imatinib is considerably more potent in inducing apoptosis of EOL-1 cells (IC₅₀ = 0.5 nM) than it is in inhibiting the phosphorylation of the fusion protein (IC₅₀ = 30 nM) (Fig. 5A). When the phosphotyrosine blots were reprobed with an antibody against PDGFRα, multiple bands were visible. These bands likely represent proteins resulting from the various splice forms of Rhe-PDGFRα. Given the relatively sharp phosphotyrosine band, not all Rhe-PDGFRα proteins derived from different splice forms may be phosphorylated.

Cell-Transforming Potential of Rhe-PDGFRα Mutationally activated tyrosine kinases, as found in many cancers, can transform the murine myeloid cell line BaF3 to interleukin-3 independence. To determine whether Rhe-PDGFRα has the ability to transform cells, a BaF3 cell line was established. BaF3 cells expressing Rhe-PDGFRα from EOL-1 were found to be IL-3-independent. The fusion protein in these cells was constitutively phosphorylated, and the phosphorylation was inhibited by imatinib with an IC₅₀ of 30 nM, the same value as obtained in EOL-1 (Fig. 5B). Inhibition with imatinib, vatalanib, and THRX-165724 resulted in reduced viability of the BaF3 Rhe-PDGFRα cells, with IC₅₀s similar to the potency of the drugs in EOL-1 (Fig. 6A). The effect of the inhibitors was overcome in the presence of IL-3 (Fig. 6B).

Discussion

We have identified and characterized a fusion gene and derived fusion kinase protein associated with idiopathic hypereosinophilia by using the EOL-1 cell line. This cell line is derived from a HES patient who had developed AML. Patients with HES are known to develop clonal cytogenic abnormalities, AML, or granulocytic sarcoma (1). We found that the viability of EOL-1 cells was exquisitely sensitive to imatinib, with an IC₅₀ of only 0.5 nM in an MTT viability assay. This finding is consistent with clinical reports of the potency of imatinib in treating HES (2–5). We also noticed that EOL-1 cells were sensitive to two other tyrosine kinase inhibitors, vatalanib and THRX-165724, albeit at higher concentrations.

EOL-1 cells were found to express a 110-kDa phosphoprotein, the phosphorylation of which was inhibited when the cells were treated with imatinib. Because constitutively activated mutant kinases are often highly phosphorylated, we isolated the 110-kDa phosphoprotein. Mass spectrometric analysis of tryptic peptides revealed peptides matching sequences predicted for the N-terminal segment of the protein encoded by NM_030917 and peptides matching sequences from the C-terminal segment of PDGFRα. Because NM_030917 and PDGFRα are located in close proximity on chromosome 4, we postulated that a chromosomal rearrangement, such as a small interstitial deletion, could have produced a fusion of these two genes without being detectable by karyotype analysis. Amplification, cloning, and sequencing of cDNAs encoding four splice variants of such a fusion from EOL-1 cells confirmed this deletion. NM_030917-PDGFRα fusion genes were also found in the blood samples from two HES patients. We suggest naming the gene coding for NM_030917 “Rearranged in hypereosinophilia,” or Rhe. To our knowledge, Rhe-PDGFRα is the first fusion kinase derived from an intrachromosomal rearrangement rather than a chromosomal translocation.

Multiple lines of evidence support the assignment of the 110-kDa phosphoprotein in EOL-1 as Rhe-PDGFRα. Based on the Rhe-PDGFRα cDNA, the corresponding protein has 849 amino acids with a predicted molecular weight of 95 kDa. This correlates well with the observed apparent molecular weight of 110 kDa for the phosphoprotein. A tryptic peptide derived from the isolated 110-kDa protein spans the predicted fusion junction of Rhe and PDGFRα. Furthermore, the 110-kDa phosphoprotein could be immunoprecipitated with an antibody against the cytoplasmic domain of PDGFRα. Rhe-PDGFRα is also likely to be the target for imatinib, vatalanib, and THRX-165724 in EOL-1 because the expression of the fusion gene conferred IL-3-independent growth to BaF3 cells that was inhibited by the three drugs at concentrations similar to those at which they inhibited the viability of EOL-1 cells. The viability of these BaF3 cells in the presence of the PDGFRα inhibitors could be maintained by exogenous IL-3.

The potencies of vatalanib and THRX-165724 in the viability and autophosphorylation assays are well correlated. In contrast, the viability of EOL-1 and BaF3 Rhe-PDGFRα cells is 100-fold more sensitive to imatinib than the inhibition of Rhe-PDGFRα autophosphorylation. The basis for this discrepancy is unknown; however, it might reflect differences in the rates of association of the different inhibitors with Rhe-PDGFRα.

Amplification and sequencing of genomic DNA from EOL-1 cells revealed that the genomic breakpoint junctions fell within an intron following exon 11 of Rhe and within exon 12 of PDGFRα. The same mutation was observed in patient 3, and a similar submicroscopic deletion was discovered in patient 1. In patient 1 the resulting transcript fused a different site in the Rhe gene to a distinct site in PDGFRα exon 12. Exon 12 encompasses the cytoplasmic juxtamembrane region of PDGFRα, followed by the kinase domain.

Patient 1 relapsed and died after initially having shown a complete remission in response to imatinib. At the time of relapse, this patient had a T674I mutation in the PDGFRα kinase domain. T674 in PDGFRα corresponds to T315 in c-Abl. Based on the crystal structure of the catalytic domain of c-Abl bound to a derivative of imatinib, T315 forms part of the imatinib binding pocket and establishes a hydrogen bond with the drug (16). The T315I mutation ablates the kinase inhibitory activity of imatinib for Bcr-Abl and is one of the most common mutations found in chronic myeloid leukemia (CML) patients who are resistant to the drug (17, 18). We speculate, therefore, that the T674I mutation underlies the relapse of patient 1. If correct, this would provide further evidence that Rhe-PDGFRα is the target of imatinib. However, we lack a cDNA sample from this patient at the start of treatment that would allow us to test in the most relevant way whether the mutation is linked to resistance by becoming predominant in HES cells at the time of relapse.

Two reports describe patients with myeloproliferative disorders with eosinophilia who have chromosomal translocations at 4q11–12 (19, 20). These translocations may involve either Rhe or PDGFRα, leading to different disease-promoting fusion proteins. For example, Rhe may play a role analogous to that played by Tel or Bcr: promoting dimerization and, thus, the activation of known oncogenic fusion kinases. Like PDGFRα, the c-Kit gene is located on chromosome 4q12. Submicroscopic deletions could also result in Rhe-c-Kit fusions. A search of the NCBI EST database reveals that Rhe is expressed in many tissues and organs, suggesting that Rhe-fusion kinases may not be restricted to cells of hematological origin. The protein encoded by Rhe is homologous (26% over 307 residues) to yeast protein FIP1, a component of a polyadenylation factor (21).

PDGFRα shares a high degree of homology with PDGFRβ (22). PDGFRβ is a target for various chromosomal translocations in chronic myeloproliferative diseases that result in the expression of fusion kinases, e.g., Tel-PDGFRβ and Rab5-PDGFRβ (23–28). Like PDGFRβ, PDGFRα has recently been described as part of a fusion gene. Baxter et al. (29) identified two atypical CML patients with a t(4, 22)(q12:q11) translocation resulting in Bcr-PDGFRα fusions. Similar to the Rhe-PDGFRα fusion we have discovered, both Bcr-PDGFRα fusions involve translocation into exon 12 of PDGFRα. Indeed, one of the two Bcr-PDGFRα fusions produces precisely the same PDGFRα fragment that we observed in Rhe-PDGFRα from EOL-1 cells.

In summary, we have shown that the Rhe-PDGFRα genomic rearrangement, discovered in the eosinophilic EOL-1 cell line, is present in a subset of patients diagnosed with HES. Cell viability and phosphorylation data suggest that the Rhe-PDGFRα kinase encoded by the fusion gene plays a central role in the disease process of these HES patients. A large number of HES/CEL samples will have to be analyzed to establish the frequency of the fusion tyrosine kinase and its general relevance to the disease. We believe that HES in which the Rhe-PDGFRα fusion is detected could more accurately be described as CEL.

Note. While this paper was under review, Cools et al. (30) described a discovery of the same 4q12 deletion in patients with HES that we have described here. The authors named NM_030917 Fip1L1 and the GenBank entry for NM_030917 has been annotated accordingly. Hence, genetic fusions between NM_030917 and PDGFRα should be referred to as Fip1L1-PDGFRα.

Supplementary Material

Acknowledgments

We thank Steve Coutre, Jason Gotlib, and Stanley Schrier (Stanford University) for providing HES patient samples, and Susan Whitman (Ohio State University) for a careful review of the manuscript. This work was supported in part by National Institutes of Health Grants CA89341 (to M.A.C.) and T32CA09338 (to R.J.B.).

References

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.