Fusion of H4/D10S170 to the Platelet-derived Growth Factor Receptor β in BCR-ABL-negative Myeloproliferative Disorders with a t(5;10)(q33;q21)¹

In approximately 95% of cases, leukocytes from patients with CML³are characterized by the presence of the BCR-ABL fusion gene, which is usually visible cytogenetically as the t(9;22)(q34;q11). The remaining 5% of patients have no evidence of either the t(9;22) or BCR-ABL. Whereas a small number of these cases appear to be indistinguishable from BCR-ABL-positive patients in terms of clinical and laboratory findings, the majority present with atypical, heterogeneous clinical features and have a relatively poor prognosis compared with BCR-ABL-positive cases. The precise classification of these aCML patients is the subject of some controversy, but it is generally held that aCML comprises part of a spectrum of MPDs and MDSs that includes CMML, polycythemia rubra vera, essential thrombocythemia, idiopathic myelofibrosis, and other cases with minimal or no granulocytic dysplasia that have usually been diagnosed as having an unclassifiable,atypical, or BCR-ABL-negative MPD (1, 2, 3).

The molecular pathogenesis of aCML, CMML, and other BCR-ABL-negative MPDs is poorly understood, principally because most cases present with a normal karyotype or have only numerical chromosomal changes. However,occasional patients are characterized by the presence of an acquired reciprocal chromosomal translocation. At the molecular level, almost all of these translocations that have been cloned thus far result in the disruption and constitutive activation of protein tyrosine kinases such as ABL, JAK2, PDGFβR, or FGFR1(4, 5, 6, 7, 8). The molecular definition of additional cases will not only lead to a better understanding of this heterogeneous group of diseases but is also likely to lead to the recognition of novel,discrete clinical entities such as the 8p11 myeloproliferative syndrome, which is specifically associated with translocations that disrupt FGFR1 (8, 9).

Here we describe a patient with a BCR-ABL-negative MPD who presented with a t(5;10)(q33;q21.2) and clinical features suggestive of CML in accelerated phase. We have cloned this translocation and demonstrated that it results in the fusion of a known gene, H4/D10S170,to PDGFβR. This is the fourth translocation involving PDGFβR described thus far, and our findings strengthen the association between deregulated tyrosine kinases and MPDs.

Patient PD, a 48-year-old man, presented with severe upper abdominal pain. This was preceded by a 6-month history of left upper quadrant and widespread skeletal pain, weight loss, and easy bruising. Clinical examination revealed massive splenomegaly palpable 20 cm below the left costal margin. The blood count showed 114 g/liter hemoglobin, a WBC count of 94 × 10⁹/liter, and 70 × 10⁹/liter platelets. The blood film showed a leukoerythroblastic picture with teardrop poikilocytes. The proportion of eosinophils was elevated at 7%, but there was no increase in blasts, monocytes, or basophils. Attempts at bone marrow aspiration yielded a dry tap. The bone marrow trephine showed granulocytic hyperplasia, established myelofibrosis (grade III-IV reticulin), and no evidence of blastic transformation. Karyotypic analysis on peripheral blood showed the presence of a t(5;10)(q33;q21.2); BCR-ABL transcripts were not detected by multiplex RT-PCR (10). A diagnosis of BCR-ABL-negative MPD (with some clinical features of CML in accelerated phase) was made, and the patient commenced on hydroxyurea with good control of symptoms. He subsequently underwent splenectomy (histology of the spleen revealed red pulp consistent with extramedullary hemopoiesis) and allogeneic bone marrow transplantation from his HLA-identical sister.

YAC clones were obtained from the Medical Research Council Human Genome Mapping Project Resource Center (Hinxton, United Kingdom). Cosmids 9-4,14C2, and 4-1 were provided by Dr. P. Marynen (Center for Human Genetics, University of Lueven, Leuven, Belgium). YAC and cosmid DNA was isolated by standard procedures and labeled with biotin by nick translation. After testing on metaphases from phytohemagglutinin-stimulated peripheral blood lymphocytes from a normal individual, labeled probes were hybridized to patient metaphases as described previously (11). In some cases, probes were cohybridized with a Cy3-labeled chromosome 10 painting probe (Cambio,Cambridge, United Kingdom). Hybridization signals were detected using FITC avidin (Vector, Peterborough, United Kingdom). Chromosomes were counterstained with 4′,6-diamidino-2-phenylindole/antifade (Biovation,Aberdeen, United Kingdom) and examined using an Olympus Vanox microscope. Images were captured using a charge-coupled device camera and SmartCapture Software (Vysis, Richmond, United Kingdom).

RNA was extracted from fresh peripheral blood mononuclear cells using the RNeasy Mini system (Qiagen, Hilden, Germany). 5′-RACE PCR was performed using a commercial kit (Life Technologies, Inc., Paisley,United Kingdom) according to the manufacturer’s instructions. Briefly,2 μg of patient RNA were reverse-transcribed using primer PDGFR-D and SuperScript reverse transcriptase. Excess nucleotides were removed by spin cartridge purification, and the cDNA was dC-tailed using terminal deoxytransferase. Tailed cDNA was amplified by a two-step heminested PCR using primers AAP + PDGFR-F in the first step and primers AAP + PDGFR-C in the second step, with 30 cycles for each reaction. Amplified products were cloned using an Original TA Cloning Kit (Invitrogen, Leek, the Netherlands) and sequenced.

Patient RNA and control RNA were reverse-transcribed as described previously (10). All primer sequences are given in Table 1. Restriction enzyme digestion and Southern blot analysis were performed under standard conditions. A PDGFβR cDNA probe(provided by Dr. A. Chantry, Imperial College School of Medicine,London, United Kingdom) that spanned the predicted breakpoint region was amplified from a full-length cDNA clone using primers PDGFR-D and PDGFR-E and gel-purified. Control DNA and RNA were extracted from patients with typical BCR-ABL-positive CML.

Standard cytogenetic analysis on peripheral blood showed that 16 of 24 metaphases harbored a t(5;10)(q33;q21.2), whereas the remaining 8 metaphases were normal. Multiplex RT-PCR analysis revealed an absence of BCR-ABL transcripts, thus excluding a diagnosis of CML (data not shown).

Chromosome band 5q33 contains PDGFβR, the receptor tyrosine kinase that is a target of the t(5;7), t(5;12), and t(5;14) in myeloid disorders (6, 12, 13). To determine whether this gene might also be disrupted by the t(5;10), we performed Southern analysis with a PDGFβR cDNA probe. Novel rearranged bands were seen in patient but not control DNA after digestion with several restriction enzymes (Fig. 1), indicating that this gene is indeed targeted by the t(5;10).

Unexpectedly, cosmids 14C2 and 4-1, which contain the 3′ and 5′ parts of the PDGFβR locus,respectively,⁴each hybridized to the der(5) and not to the der(10) (Fig. 2). If a simple reciprocal translocation had taken place, it would have been anticipated that either one cosmid would hybridize to both derivative chromosomes or that 14C2 would hybridize to the der(5), and 4-1 would hybridize to the der(10). YACs 745d10 and 756a2, which flank but do not contain PDGFβR (14),also both hybridized to the der(5) and not the der(10). Further analysis demonstrated that YAC 816d6, which is further distal to PDGFβR (15), hybridized to both the der(10) and the der(5), whereas YAC 850c10 hybridized to the der(10) only. This suggested that a complex rearrangement had taken place, with the first event having been a translocation between the region spanned by YAC 816d6 and chromosome 10, and the second event having been an inversion on the der(5) resulting in a fusion of H4 and PDGFβR (Fig. 3,A). This is likely to have taken place as a single event involving four chromosomal breaks (Fig. 3,B). This model predicts that YAC 756a2 should lie in between cosmids 9-4 and 4-1 on the der(5) but be distal to the two cosmids on normal chromosome 5. These orders were confirmed by fiber FISH: on normal copies of chromosome 5, the YAC 756a2 signal was seen to partially overlap with the two cosmids, whereas on the der(5), YAC 756a2 is flanked by cosmid signals (Fig. 2, E and F). It is likely that YAC 756a6 is also partially deleted on the der(5) because the signal from this clone consistently and almost completely overlapped with the cosmid signals. Complex rearrangements appear to be a common feature of rare translocations that in many cases appear karyotypically to be straightforward genetic exchanges, for example, the ZNF198-FGFR1,TEL-ABL, and AF10-MLL fusions that result from the t(8;13), t(9;12),and t(10;11), respectively (Ref. 8 and the references therein). The data presented here serve as an illustration that FISH analysis may give misleading results in the analysis of chromosomal translocations. In this case, for example, FISH analysis alone would have suggested that PDGFβR was not involved in the t(5;10).

To determine the site of breakage on chromosome 10, FISH was performed with an ordered series of YAC clones that were known to map to 10q11–22. On patient metaphases, YACs 794g4, 942d3, 845g11, 916d5, and 928d9 hybridized to the der(10), whereas YACs 954a2, 948h12, and 750h4 hybridized to the der(5). All clones hybridized to normal chromosome 10, and some also hybridized to other chromosomes in some cases due to chimerism. These data indicated a breakpoint between YACs 928d9 and 954a2 in the vicinity of marker D10S207 (15). Subsequent analysis showed that YAC 940e4 hybridized to both derivative chromosomes and therefore spanned the chromosome 10 breakpoint (Figs. 1,C and 4).

All cases that have been analyzed thus far with a t(5;7), t(5;12), or t(5;14) result in the fusion of distinct partner genes to a common exon of PDGFβR (6, 12, 13). To determine whether the t(5;10) resulted in a novel fusion gene, we used 5′-RACE PCR using primers within or downstream of this common PDGFβR exon. Several clones were isolated in which the sequence diverged from PDGFβR at precisely the same point as that seen in the other three translocations. A BLAST search revealed that the novel sequence was derived from a known gene, H4/D10S170 (16). PCR analysis demonstrated that this gene was contained in YAC 940e4(not shown).

To confirm the presence of chimeric mRNA in patient leukocytes, we performed RT-PCR analysis. H4-PDGFβR chimeric mRNA was detected specifically in the patient RNA but not in control RNA (Fig. 5). Reciprocal PDGFβR-H4 transcripts were not detected. The break in H4 occurred between positions 1141 and 1142 of the cDNA sequence (GenBank accession number S72869) and resulted in an in-frame fusion to PDGFβR. The H4-PDGFβR fusion gene is predicted to encode a M_r 107,000 protein of 948 amino acids that contains the leucine zipper from H4 and the entire transmembrane and tyrosine kinase domains of PDGFβR (Fig. 6).

Chromosomal translocations that target specific non-receptor and receptor tyrosine kinases are seen frequently in hematological malignancies (Table 2) and in all cases result in the NH₂-terminal fusion of a partner gene to the COOH-terminal catalytic domain of the tyrosine kinase. The resulting chimeric gene is transcribed and translated into a fusion protein. Functional analysis has demonstrated that many of these chimeric proteins, such as BCR-ABL, TEL-JAK2, and ZNF198-FGFR1, possess transforming activity as a result of the constitutive activation of their tyrosine kinase moieties(17, 18, 19). Constitutive activity arises by partner gene-dependent oligomerization of fusion proteins and thus mimics the normal process of tyrosine kinase signaling after binding of their cognate ligands (19, 20). In addition, the partner gene may relocalize the tyrosine kinase to a different cellular compartment from that in which it normally resides, thus potentially enabling the kinase to phosphorylate novel substrates. It is particularly striking that deregulated tyrosine kinases are involved in virtually all cases of MPD for which the underlying pathogenesis is known, including the case described here.

We have identified H4 as the fourth partner gene that becomes fused to PDGFβR as a result of chromosomal translocation, with the other three being HIP1, TEL/ETV6, and CEV14 (Table 2). Recently, we have become aware of an additional case with a t(5;10) and H4-PDGFβR fusion, indicating that this abnormality is recurrent (21). Of these four fusions, the TEL-PDGFβR and HIP1-PDGFβR chimeric proteins have been shown to possess constitutive tyrosine kinase activity and transforming ability(12, 22); it is highly likely that this is the case for the other two fusions as well.

H4 is a widely expressed gene that is fused to ret as a result of an inv(10)(q22q21) in a subset of papillary thyroid carcinomas (23, 24). The H4-ret fusion protein, usually referred to as PTC-1, is a constitutively active tyrosine kinase. Transgenic mice with thyroid-targeted expression of PTC-1 developed thyroid carcinomas, indicating that this fusion plays a primary role in the pathogenesis of the disease with which it is associated (25). H4 is not the only gene that is disrupted in both solid tumors and hematological malignancies: for example, TPM3, which encodes a nonmuscular tropomyosin, is fused to ALK in anaplastic large cell lymphoma with a t(1;2)and TRK in a subset of papillary thyroid carcinomas(26). Furthermore, the ETV6-TRKC fusion has been described in both patients with congenital fibrosarcoma and patients with acute myeloid leukemia (27, 28).

H4 shows weak but significant homology to the myosin superfamily, but its normal function is unclear. H4 contains a NH₂-terminal leucine zipper that mediates dimerization of PTC-1 and is essential for the constitutive activation of its tyrosine kinase (29). This leucine zipper is also present in H4-PDGFβR. In addition, the H4 promoter drives the expression of constitutively active ret in thyroid follicular cells, a tissue in which it is not usually expressed. The mechanism of transformation by H4-PDGFβR is presumably similar, except that in this case, H4 is driving constitutive PDGFβR tyrosine kinase activity in primitive hemopoietic cells.

It has been recognized for several years that the t(5;12) is associated with an unusual MPD/MDS that is difficult to classify within defined French-American-British subtypes (30, 31, 32). Specifically,patients typically present with eosinophilia plus other clinical features that are suggestive of both CML and CMML. Very similar clinical pictures were seen in other patients who had diseases characterized by primary deregulation of PDGFβR, e.g.,those associated with the t(5;7) or the t(5;10). The fourth fusion,CEV14-PDGFβR, was different in that it arose as a secondary abnormality at relapse in a patient who presented initially with AML and a t(7;11). However, it is notable that the acquisition of the t(5;14) was associated with the appearance of marked eosinophilia and hepatosplenomegaly, i.e., features usually associated with a MPD (33). Remarkably, and at the current time,inexplicably, all 27 cases that have been reported with a MPD/MDS and a t(5;12), t(5;7), or t(5;10) have been male (12, 21, 31). Taken together, these observations suggest that primary deregulation of PDGFβR leads to a consistent phenotype, which might be better designated as a specific clinical subtype.

We thank Dr. Tom Vulliamy for invaluable advice, Dr. Andrew Chantry for providing the PDGFβR cDNA clone,and Dr. Peter Marynen for providing the PDGFβRcosmid clones.