In this issue of Blood, Patel et al identify novel recurrent acquired mutations in exon 13 of the Janus kinase 2 gene (JAK2ex13indel) in 3 patients with chronic eosinophilic leukemia (CEL), 2 of whom also met the diagnostic criteria for polycythemia vera (PV).1
PV and CEL not otherwise specified are clonal hematologic disorders classified as myeloproliferative neoplasms (MPNs).2 PV is characterized by increased red cell mass and hematocrit >48%, whereas CEL presents with overproduction of eosinophils (>1500/µL) and the presence of bone marrow blasts (5% to 20%). PV is the result of an acquired JAK2V617F mutation in exon 14 in more than 95% of patients. In about 2% of patients, PV is the result of mutations in JAK2 exon 12 (JAK2ex12). Conversely, genetic causes for CEL are largely unknown. Rearrangements of the platelet-derived growth factor receptor α (PDGFRA) and β (PDGFRB) and the fibroblast growth factor receptor 1 (FGFR1) genes that define myeloid/lymphoid neoplasms with eosinophilia are excluded from CEL by diagnosis criteria.3
The differentiation process that leads to the production of red cells or eosinophils starts with the differentiation of multipotent hematopoietic stem cells into progenitors that give either erythroid or eosinophil precursors. The main cytokine driving the erythroid differentiation is erythropoietin (EPO) through the activation of the homodimeric type I receptor EPOR/JAK2/STAT5 axis.4 Meanwhile, eosinophil differentiation relies mainly on interleukin-5 (IL-5), and also on IL-3 and granulocyte-macrophage colony-stimulating factor (GM-CSF). The IL-5 heterodimeric receptor (IL-5R) shares a common β signaling chain (βc) with the receptors of IL-3 and GM-CSF. All 3 are preassociated preferentially with JAK2, which leads to activation of signal transducer and activator of transcription 5 (STAT5), phosphatidylinositol 3-kinase (PI3K)/AKT, and mitogen-activated protein kinase (MAPK) signaling pathways upon ligand binding to their specific Rα chains.5
Patel et al identified JAK2ex13indel mutations in 3 of 173 patients with CEL, including some with concomitant PV, and referenced another similar patient described by Pardanani et al.6 The JAK2ex13indel, similar to JAK2ex12, was found only in the heterozygous state in erythroid progenitors.7 In JAK2V617F-positive PV, progenitors may be homozygous. The JAK2ex13indel mutation was found at a higher allele frequency in eosinophils, suggesting that it also gives a proliferative advantage in this lineage. Thus, the JAK2/STAT5 pathway is involved in the development of eosinophilia. JAK2V617F has been found in 4% of patients with hypereosinophilic syndrome and CEL.8 Moreover, a STAT5BN642H-activating mutation was recently reported in other myeloid and myeloid/lymphoid malignancies with eosinophilia.9 Some rearrangements involving JAK2, such as PCM1-JAK2, BCR-JAK2, and ETV6-JAK2 have also been found in myeloid/lymphoid neoplasms with eosinophilia.
JAK2ex13indel leads to a 4-amino acid deletion and 1 variable amino acid insertion within the pseudokinase (JH2) domain of JAK2 (Leu583-Ala586DelInsSer/Gln/Pro). The authors showed that they are gain-of-function mutations based on mild EPO-independent growth of JAK2ex13indel-positive progenitors, IL-3–independent growth of Ba/F3 cell lines overexpressing JAK2ex13indel, and constitutive activation of JAK2/STAT5/ERK1/2 signaling pathways in the absence of any cytokine but in the presence of several cytokine receptors. In silico modeling predicts that the JAK2V617F and JAK2ex13indel use the same mechanism of activation. Indeed, genetic alteration of the negatively charged amino acid E596 to a neutral R or positively charged K, restores the autoinhibition of the JH2 domain in JAK2V617F and also in JAK2ex13indel, and it consequently disrupts their constitutive kinase activity. JAK2ex13indel-mediated activation of STAT5 is achieved in 2 ways: (1) interaction and activation of the common βc, particularly IL-5R and also IL-3R and GM-CSF receptor (GM-CSFR), suggesting that the simultaneous activation of the 3 receptors is responsible for eosinophilia, and (2) activation of homodimeric type I receptors, more particularly EPOR, which explains the PV. To a lesser extent, the thrombopoietin receptor could also be induced. Further investigation is needed to determine to what extent this receptor can be activated and whether this specific JAK2 mutation could be identified in other MPNs. In contrast, JAK2V617F more specifically activates homodimeric type I receptors, and JAK2ex12 drives EPOR and IL-3R signaling.7 It is interesting to have established that many different mutations of JAK2 can give specific phenotypes, including isolated thrombocytosis (JAK2R567Q, JAK2V617I, JAK2S755R/R938Q, JAK2R867Q) or erythrocytosis (JAK2R846Q, JAK2ex12). Using a mouse model may allow demonstration of the exact hematopoietic phenotype and mechanism induced by JAK2ex13indel compared with other JAK2 mutations.
Because very few driver mutations have been described so far in CEL, the discovery of JAK2ex13indel is very important in determining the molecular landscape associated with the main clinical presentation. These mutations will require careful examination because they are difficult to detect by next-generation sequencing. As shown by the authors, these diseases can harbor additional mutations such as in TET2 or DNMT3A occurring before or after JAK2ex13indel that could play a role in the fitness of the clone. When identifying such mutations in a clonal CEL, gain-of-function mutations in JAK2 or potentially in IL-5R should be investigated when searching for the signaling drivers of the disease. The activation of JAK2 and STAT5 could also be used to discriminate the different entities with eosinophilia. For instance, the provisional entity defined by the PCM1-JAK2 fusion protein uses activation of the MAPK signaling pathway rather than that of JAK2 and STAT5, and thus it should probably remain classified as a non-MPN molecular cause of myeloid neoplasms associated with eosinophilia.10
Although phosphorylation of JAK2/STAT5 in PCM1-JAK2 disorder is not as marked as in CEL, the use of JAK1/2 inhibitors such as ruxolitinib seems to be equally efficient in reducing eosinophilia in both diseases. Antibodies against IL-5 and IL-5R are also in active development.3 Future studies should explore the complications and prognosis of the JAK2ex13indel disorders because both eosinophilia and PV have an increased risk of thrombosis.
Conflict-of-interest disclosure: The authors declare no competing financial interests.