Molecular aspects of myeloproliferative neoplasms
- First Online:
- Cite this article as:
- Delhommeau, F., Jeziorowska, D., Marzac, C. et al. Int J Hematol (2010) 91: 165. doi:10.1007/s12185-010-0530-z
During these past 5 years several studies have provided major genetic insights into the pathogenesis of the so-called classical myeloproliferative neoplasms (MPNs): polycythemia vera (PV), essential thrombocythemia (ET), and primary myelofibrosis (PMF). The discovery of the JAK2V617F mutation first, then of the JAK2 exon 12 and MPLW515 mutations, have modified the understanding of these diseases, their diagnosis, and management. Now it is established that almost 100% of PV patients present a JAK2 mutation. Nearly 60% of ET patients and 50% of patients with PMF have the JAK2V617F mutation. The MPLW515 mutations are also present in a small proportion of ET and PMF patients. These mutations are oncogenic events that cause these disorders; however, they do not explain the heterogeneity of the entities in which they occur. Genetic defects have not been yet identified in around 40% of ET and PMF. There are likely additional somatic genetic factors important for the MPN phenotype like the recently described TET2, ASXL1, and CBL mutations. Moreover, polymorphisms in the JAK2 gene have been recently described as associated with MPN. Additional studies of large cohorts are required to dissect the genetic events in MPNs and the mechanisms of these oncogenic cooperations.
KeywordsMyeloproliferative neoplasms JAK2 MPL TET2 CBL ASXL1
Myeloproliferative neoplasms (MPNs) are clonal hematopoietic malignancies resulting from the transformation of a hematopoietic stem cell (HSC), with abnormal amplification of one or several myeloid lineages. The classic MPNs comprise chronic myelogenous leukemia (CML), polycythemia vera (PV), essential thrombocythemia (ET), and primary myelofibrosis (PMF). Less frequent MPNs are chronic neutrophilic leukemia (CNL), chronic eosinophilic leukemia (CEL), systemic mastocytosis (SM), and other unclassifiable entities. Some myeloid malignancies with myeloproliferative features, such as chronic myelo-monocytic leukemia (CMML) and atypical chronic myelogenous leukemia (aCML) are classified as mixed myelodysplastic/myeloproliferative diseases (MDS/MPN).
MPNs frequently involve the deregulation of a tyrosine kinase that mimics the signaling pathways induced by hematopoietic cytokine receptors. CML is the paradigm for the pathogenesis of MPN. It has been shown that the kinase domain of the ABL tyrosine kinase is constitutively activated in the cytoplasm, following oligomerization via the BCR coiled-coil fused domain. The inhibition of the ABL kinase activity clinically results in disease regression, demonstrating that BCR-ABL is the principal oncogenic event to be targeted in this disease.
2 The cytokine receptor and JAK2 axis
ET, PV, and PMF share various biological features, which suggested a common molecular origin. The most striking similarity is a hypersensitivity to erythropoietin (EPO), which leads to the growth of erythroid progenitors in vitro in the absence of cytokines . This process has been called “endogenous erythroid colonies” (EEC) formation. EEC are observed in 100% of PV, about 30–50% of ET and PMF. In fact, hypersensitivity is not restricted to EPO but also involves other cytokines such as interleukin-3 (IL-3), SCF, insulin-like growth factor-1 (IGF-1), granulocyte macrophage-colony stimulating factor (GM-CSF), and TPO. Interestingly, endogenous megakaryocyte colony formation is observed in most cases of MPN, including ET without EEC . Therefore, it was anticipated that the oncogenic event causing MPN would hit a molecule involved in the signaling pathways of hematopoietic cytokines, especially EPO and TPO.
In 2005, several groups, using functional, genetic, and molecular approaches, simultaneously identified JAK2V617F as the major molecular etiology of BCR/ABL-negative classic MPNs, paving the way for the subsequent discovery of other JAK2 and MPL mutations.
2.1 The JAK2V617F mutation
JAK2 is a member of a group of proteins with tyrosine kinase activity, the Janus kinases (JAK). These proteins bind the intracytoplasmic domain of various cytokine receptors via their FERM (4.1 Ezrin Radixin Moesin)-like domain, promoting downstream cell signaling . JAK2 mediates signaling by various hematopoietic receptors comprising the EPO, TPO, and granulocyte colony-stimulating factor receptors. JAK2 is also involved in signaling by receptors with tyrosine kinase activity, such as KIT. Following cytokine binding, the two JAK2 proteins bound to the cytosolic domain of the receptor are activated by transphosphorylation. In turn, they phosphorylate the tyrosines of the receptor. This induces the recruitment and phosphorylation of actors of signal transduction pathways, such as PI3K, RAS complex, and STAT5a/b. Thus, JAK2 mediates cytokine signaling to regulate cell proliferation, differentiation, and anti-apoptotic events. JAK2 also promotes maturation and is required for the efficient trafficking of homodimeric type 1 receptors, such as EPOR  and MPL . JAK2 stabilizes the mature form of MPL and promotes its recycling. These two processes are defective in MPNs .
For the mutant cells, a way to express higher levels of JAK2V617F is to duplicate the mutant allele. This duplication is achieved in malignant cells from a large proportion of MPN patients. HSC with one JAK2V617F allele, also called heterozygous JAK2wild-type/V617F cells, undergo mitotic recombination on the short arm of chromosome 9 where the JAK2 locus lands . By this way, they give rise to a homozygous JAK2V617F subclone with two mutant alleles. Intriguingly, this phenomenon occurs in some PMF patients and virtually all patients with PV but none with ET, as demonstrated by the genotyping of individual colonies grown from myeloid progenitors [14, 15].
An impressive number of studies allow establishing the frequency of the JAK2V617F mutation in MPN. The mutation has been found in more than 95% of PV, around 60% of ET, and 50% of PMF cases . In addition, some cases of other myeloid malignancies, such as CNL, uMPN, CMML, refractory anemia with ring sideroblasts and thrombocytosis, and rare AML have this mutation.
All these observations have greatly improved the understanding of the pathophysiology of MPN. However, although JAK2V617F positive PV, ET, and PMF display differences in terms of clonal architecture, JAK2V617F allele burden, and HSC behavior, the reason why this unique mutation is present in such distinct entities remains to be fully elucidated.
2.2 JAK2 exon 12 mutations
In the rare JAK2V617F negative PV, Scott et al.  described four somatic gain of function mutations involving residues 538–543 within exon 12. These mutations span a region between the SH2 and JH2 domains (Fig. 1). They include a point mutation (K539L), a double mutation (H538QK539L), a 2-amino acid deletion (N542-E543del), and a 2-amino acid deletion followed by an insertion (F537-K539delinsL). Other mutations in exon 12 have also been described and 15 different acquired mutations in the SH2 domain of JAK2 have been published . All the defects in JAK2 exon 12 lead to a gain of protein functions that confer growth factor independency when expressed in the Ba/F3-EPO-R cell line (Fig. 2). Clinically, these mutations are associated with the presence of EEC and a low EPO plasma level. Patients with somatic activating missense or deletion mutations in exon 12 most commonly present an isolated erythrocytosis but not the pan-myeloid expansion seen in JAK2V617F positive PV. Therefore, the clinical features of these rare PV are close to idiopathic erythrocytosis. The reasons why the diseases induced by JAK2V617F and JAK2 exon 12 mutations are phenotypically different remains unknown. One hypothesis is that the exon 12 mutants induce a higher constitutive kinase activity than JAK2V617F and induces polycythemia rather than a thrombocytosis. Another hypothesis is that JAK2 exon 12 mutant proteins interact preferentially with EPO-R.
2.3 MPL mutations
In 2006, in two studies samples from JAK2V617F negative ET or MF patients were examined for mutations in the cytokine receptors that bind JAK2 [19, 20]. The authors described a substitution of a somatic tryptophan to leucine/lysine in position 515 on the TPO receptor MPL (MPLW515L/K). Further studies led to the discovery of other mutations in the same codon (MPLW515A)  (Fig. 1). These mutations are present in 5–10% of patients with primary MF and 2–5% of patients with ET .
The 515 amino acid is located in a stretch of five amino acids (K/RWQFP) found just after the transmembrane domain. These five amino acids play a major role in the cytosolic conformation of MPL and prevent spontaneous activation of the receptor. Mutations of W or K (mouse)/R (human) are sufficient to activate the receptor  (Fig. 2). As observed for JAK2V617F, these mutations might be present in one or two copies. MPLW515 mutants target a HSC capable of reconstituting NOD/SCID mice, demonstrating that mutations occur in HSC . Patients with MPLW515 mutations are characterized by spontaneous MK growth with absence of EEC . In a murine model, expression of MPLW515L leads to a lethal myeloproliferative disorder with thrombocytosis and myelofibrosis that is dependent on the JAK2 activation . Recently, it has been shown that a MPLS505N mutation, initially described in familial ET, could be also found in sporadic ET or PMF .
2.4 C-CBL mutations
C-CBL is an E3 ubiquitin ligase that negatively regulates signal transduction of activated tyrosine kinases. CBL acts by the ubiquitination of various growth factor receptors, which results in their internalization and degradation . In 2009, acquired uniparental disomies (aUPD) of the long arm of chromosome 11 spanning the C-CBL gene were described. CBL mutations were then identified in 17% of CMML and in <10% of PMF cases [26, 27] (Fig. 1). Subsequently, mutations not only in C-CBL, but also in CBL-B, were found in cases of AML transformations of MPN, including CML [28, 29]. These data indicate that C-CBL mutations can occur in PMF or during the progression of MPN to AML. As there was a strong association of C-CBL mutations with aUPD and loss of the wild-type allele, these defects were expected to be loss of function mutations of a tumor suppressor, with an impaired down-regulation of tyrosine kinase signaling cascades downstream of growth factor receptors. Surprisingly, a gain-of-function effect was found when introducing C-CBL mutants in c-Cbl−/− mouse hematopoietic stem/progenitor cells. This hypersensitivity to various cytokines, including SCF, TPO, FLT3-ligand, and IL3 was reduced by the presence of wild-type C-CBL, suggesting that the loss of wild-type functional CBL is necessary for the pathogenic effect of CBL mutants in myeloid malignancies  (Fig. 2). These findings identify CBL as an intriguing ambivalent oncogene and tumor suppressor gene. Its precise role in the pathogenesis of MPN remains to be elucidated .
3 Inherited predisposition and putative initiating events
Most mutations or DNA structural rearrangements in MPNs result in a constitutive (or enhanced) tyrosine kinase or receptor signaling activation. However, a series of observations led to the hypothesis that all these lesions were insufficient for the onset of MPNs (Figs. 1, 2). Other malignancies involve the constitutive activation of tyrosine kinases, making these oncogenic mechanisms not specific to MPN. In addition, striking differences exist between PV, ET, and PMF, which all share the JAK2V617F mutation. To explain these differences, the first hypothesis was that a distinct target cell could initiate different diseases. This hypothesis was rapidly ruled out by the demonstration that the target cell of the JAK2V617F was the HSC in ET, PV, and PMF [21, 24, 31]. The second hypothesis proposed that variation of the degree of activation of the JAK2 tyrosine kinase would result in distinct diseases. A low JAK2V617F signaling would lead to ET, whereas increased constitutive signaling would result in PV or PMF. This has been clearly shown both in primary cells from patients and in mouse models. First, the ET patients have lower JAK2V617F allele burden than PV and PMF patients. Second, Epo sensitivity of homozygous progenitors with two mutant alleles was demonstrated to be bigger than that of heterozygous JAK2wild-type/V617F progenitors in vitro . Finally, studies of retroviral transfer models or transgenic mice models have presented a thrombocythemic phenotype in case of low JAK2V617F expression that was in striking contrast with a phenotype of PV leading to MF in case of high JAK2V617F expression [32, 33, 34, 35, 36]. The third hypothesis, which claims that other germline or acquired molecular lesions could modify the phenotype of the disease, has been considered as highly probable with growing evidences. In mouse models, phenotypic differences were found according to mouse strains, supporting the hypothesis that a distinct genetic background could result in distinct diseases carrying the same acquired JAK2V617F mutation . In addition, clonality and cytogenetic studies had shown that in some MPN patients, the size of the malignant clone as determined either by X inactivation or by FISH detection of del (20q), was obviously bigger than one would have expected according to the quantification of JAK2V617F alleles [37, 38, 39]. Finally, secondary acute myeloid leukemia arising from a JAK2V617F MPN was found to be predominantly JAK2wild type . These data led to a model where a HSC carrying a pre-JAK2 molecular defect (either germline or acquired) would be the target of secondary oncogenic molecular lesions able to trigger the myeloproliferative phenotype.
Many groups have searched for pre-JAK2 molecular events in MPN. This led to the discovery of TET2 (TET oncogene family member-2) and ASXL1 (additional sex comb like-1) defects and to the identification of the 46/1 predisposition haplotype in JAK2.
3.1 Susceptibility to MPN due to the JAK2 46/1 haplotype
An inherited predisposition was suspected since the observation of pedigrees with many MPN affected family members with or without acquired JAK2 mutations . In addition to these obvious familial MPN, large population studies found that first-degree relatives of affected patients had a five- to sevenfold increased risk for developing an MPN . In 2009, an intriguing association between the risk of developing a JAK2V617F MPN and a germline haplotype block spanning the JAK2 gene was described by three independent groups [43, 44, 45]. This predisposition allele is called “46/1” and is tagged by four single nucleotide polymorphisms (SNP) on the 3′ end of the JAK2 gene (rs3780367, rs10974944, rs12343867 and rs1159782). The analysis of patients heterozygous for the 46/1 haplotype allowed the demonstration that the JAK2 V617F mutation occurs more likely in cis with the predisposition allele. This observation led to the hypothesis that this particular haplotype could confer an unexplained hypermutability property on the JAK2 locus. In line with this hypothesis, JAK2 exon 12 mutations were also demonstrated to mainly occur in cis with 46/1. However, a second hypothesis proposes that the JAK2V617F mutation occurs at the same frequency on the two alleles, but cells with the mutation on the predisposition allele have a stronger selective advantage. This so-called “fertile ground” hypothesis may be explained by a functional variant of JAK2, but there is no allele-specific differential expression of JAK2, and the 46/1 haplotype is not linked with any nonsynonymous coding polymorphism. Things are probably more complex than expected because JAK2 negative MPN are also associated with the predisposition haplotype, although at a lower extent . This observation would favor the “fertile ground” hypothesis, assuming that a subtle proliferative advantage could preexist in cells with the haplotype predisposition and no JAK2 mutation. The two hypotheses are currently under investigation. Perhaps, a solution to this conundrum will be a mix of the two explanations, with the fertile ground hyperproliferation indirectly leading to an increased frequency of error-prone mutations, which may be more likely to occur on a hypermutable allele.
3.2 TET2 mutations
TET2 belongs to a family of three genes. The founder member, TET1 (Ten eleven translocation-1), is involved in a t(10;11) translocation in rare acute myeloid leukemia . TET1 is a 2-oxoglutarate and Fe2+ dependent dioxygenase that catalyzes the conversion of 5-methylcytosine (5mC) to 5-hydroxymethylcytosine (5hmC) in embryonic stem cells . The three members of the TET family (TET1, TET2, and TET3) share two highly conserved domains. The carboxy-terminal domain holds the catalytic dioxygenase activity. The conservation of this domain in the family suggests that TET2 and TET3 are also dioxygenases able to convert 5mC to 5hmC (Fig. 2). The meaning of the conversion of 5mC to 5hmC remains unclear. However, its putative function is to participate in an active or passive DNA demethylation process.
TET2 mutations and deletions have recently been identified in MPN . Initial observations were done by analyzing a subset of MPN patients who had an uncommon dynamics of the JAK2V617F malignant clone. In striking contrast with other JAK2V617F MPN patients, who have a late amplification of the clone related to the effect of JAK2V617F, a minority of patients had an early onset of clonality. This was due to the amplification of the JAK2V617F clone in early myeloid or multipotent compartment of progenitor cells. SNP arrays and CGH array revealed in three out of five such patients that the TET2 locus was rearranged either in an acquired 325 kb 4q24 deletion or in an acquired uniparental disomy of chromosome 4q. Sequencing of the coding sequence of TET2 revealed acquired mutations in the two patients with aUPD. Subsequent analysis of TET2 sequence and rearrangements in more than 1000 patients by several groups led to the establishment of the frequencies of TET2 defects in MPN: around 12% in all MPN [28, 50, 51], 14% in PV, 8% in ET, 20% in PMF, 29% in SM , and 25% in post-MPN-AML [28, 50, 51]. TET2 defects were detected in MPN carrying JAK2V617F, JAK2 exon 12 mutations, MPL515 mutations, or devoid of these mutations. Studies on familial MPN found only acquired mutations in TET2 . However, a recent study described two siblings, one with a JAK2V617F PV, and the other without hematopoietic malignancy, presenting a germline frameshift mutation in TET2 .
In contrast to JAK2 and MPL mutations, TET2 defects are far to be specific for MPN. Rearrangements at the TET2 locus had already been observed in MDS and AML , and TET2 alterations have been detected in around 20% of myelodysplastic syndromes, 15% of acute myeloid leukemia, and 40% of CMML [49, 56, 57, 58].
Besides whole gene deletions, most of the molecular defects found in the TET2 sequence result in the putative alteration of the function of the protein (Fig. 2). The majority of the mutations are nonsense mutations or small insertions or deletions leading to a frame shift, all encoding truncated proteins. Some patients have missense mutations localized in the two highly conserved domains that presumably also alter the function of the protein. In a small number of cases, acquired missense mutations outside of the conserved domains and splicing site mutations have been found. In 80% of the cases of MPN with TET2 mutation, the defect involves one copy of the gene. Thus, TET2 haploinsufficiency may alter hematopoietic cell functions in most patients. However, in a minority of patients, two defects, sometimes as the result of mitotic recombination, are found [49, 54, 55, 56, 57, 58]. This suggests that the loss of the two functional copies of TET2 may participate in the initiation or the progression of these diseases. The analysis of serial samples and individual colonies from patients with JAK2 and TET2 mutations showed that TET2 mutations are a pre-JAK2 event in the majority of the cases . However, in some patients, JAK2 mutations precede TET2 mutations, and in rare cases, a biclonal disease is detected with independent JAK2 and TET2 clones . In conclusion, most TET2 mutations are acquired, can precede or not JAK2 mutations, may occur in distinct pathological clones, and infrequently target the two copies of TET2.
The pathogenic role of TET2 is currently unknown. However, one can speculate that TET2 inactivation may have epigenetic consequences that deregulate genes involved in early hematopoiesis as well as in myeloid differentiation . A current hypothesis is that TET2 mutations can both participate in the initiation of a pre-malignant clone and the progression of the disease.
3.3 ASXL1 mutations
ASXL1 belongs to a three-member family of enhancers of trithorax and polycomb proteins (ASXL1, 2, 3) involved in the maintenance of activation and the silencing of development-related genes through chromatin remodeling [60, 61]. The ASXL genes share conserved domains: an amino-terminal ASX homology (ASXH) domain and a C-terminal plant homeodomain (PHD) finger. ASXL1 is part of a repressive complex containing histone H1.2, and has an ambivalent transcriptional regulator role in retinoic acid receptor-mediated transcription through its association with the heterochromatin protein 1 (HP1) and the H3 demethylase LSD1 [60, 61]. The hematopoietic function of ASXL1 is still unclear: the knock-out mouse model led to mild hematopoietic phenotypes and failed to induce hematopoietic malignancies, suggesting that other ASXL proteins may have redundant functions .
Mutations in ASXL1 were found in 10% of myelodysplastic syndromes and 40% of CMML after the identification by SNP arrays and CGH array of recurrent deletions or UPD at chromosome 20q . ASXL1 was subsequently sequenced in a series of MPN where it was found mutated in 15% of cases and in around 20% of transformations of MPN to AML [28, 64]. ASXL1 mutations are mostly located in exon 12, result in a truncated protein lacking the PHD domain (frameshift or nonsense mutations) (Fig. 2), and are heterozygous. These data suggest that haploinsufficiency of ASXL1 may have a role in the pathogenesis of MPN and other myeloid malignancies. Analysis of serial samples from patients with MPN who evolved into AML demonstrated that ASXL1 mutations were always present at the chronic phase of the disease .
FD is a recipient of a new investigator grant from the MPD foundation, and of grants from Association Laurette Fugain, foundation de France, and Inca-DHOS Ile de France.