Epigenetics and mutations in chronic myeloproliferative neoplasms

The BCR-ABL1-negative classic myeloproliferative neoplasms, polycythemia vera (PV), essential thrombocytemia (ET) and primary myelofibrosis are clonal stem cell disorders associated with an increased production of mature blood cells belonging preferentially to one cell linage.1 They share substantial phenotypic mimicry, can undergo phenotypic shifts (from PV to ET and vice versa) as well as evolution to myelofibrosis (post-PV/post-ET myelofibrosis), and all eventually progress to leukemia. The hypothesis that hypersensitivity of hematopoietic stem and progenitor cells to cytokines might largely account for the pathogenesis of myeloproliferative neoplasms has been corroborated by the discovery of mutations that affect cytoplasmic proteins involved in cytokine signaling, either resulting in a gain-of-function (JAK2 and MPL) or a loss-of-function (CBL and LNK). Dysregulation of tyrosine kinases is a recurrent theme in chronic myeloid neoplasms, as exemplified by the constitutive activation of ABL caused by oligomerization of the BCR-ABL fusion protein in chronic myelogenous leukemia, the gain-of-function mutation of the tyrosine kinase receptor c-KIT in mastocytosis, and the activation of platelet-derived growth factor receptor-α or -β and fibroblast growth factor receptor in hypereosinophilic disorders. However, high-throughput genomic analyses of myeloproliferative neoplasms have recently identified a second group of mutations that affect proteins involved in the epigenetic regulation of transcription, such as TET2, ASXL1 and EZH2.2 These abnormalities can occur in association and/or with mutations targeting tyrosine kinases. However, unlike the JAK2V617F, JAK2 exon-12 and MPLW515 mutations, which have been identified very rarely outside the classic myeloproliferative neoplasms, TET2, ASXL1 and EZH2 are mutated in a wide spectrum of myeloid malignancies including myelodysplastic syndromes, myelodysplastic syndromes/myeloproliferative neoplasms and acute myeloid leukemias, suggesting that these mutations might contribute a common genomic hit in myeloid malignancies. Abnormalities in other epigenetic regulators, due to mutations in IDH1 and IDH2 and DNMT3A, have been detected preferentially in association with leukemic transformation of chronic myeloproliferative neoplasms as well as in de-novo leukemias.

Figure 1.Known functions of EZH2, TET2, ASXL1 and JAK2 in epigenetic regulation of gene expression. The upper part of each box schematically depicts the main known function of each protein, while the effects of abnormal proteins are shown in the lower part. TET2 normally converts 5-methylcytosine (5mC) to 5-hydroxymethylcytosines (5hmC). EZH2 is the catalytic subunit of the PRC2 complex and trimethylates Lys-27 of histone H3 (H3K27) leading to transcriptional repression of target genes. ASXL1 exists in chromatin-associated multiprotein complexes, together with PcG and TrxG proteins, involved in modifications of chromatin configuration that result in repressed and enhanced transcription, respectively, in a cellular context-specific manner. Described loss-of-function mutations of EZH2, TET2 and ASXL1 presumably lead to suppression of catalytic activity of these enzymes. Mutant JAK2, but not the wild-type protein, phosphorylates protein arginine methyltransferase 5 (PRMT5), causing inhibition of its arginine methyltransferase activity on H2A and H4 (H2AR3me and H4R3me). JAK2 also phosphorylates Tyr 41 (Y41) on histone H3 leading to decreased HP1α binding to chromatin; the displacement of HP1α is magnified after enhanced H3Y41 phosphorylation due to JAK2V617F.

Table 1.ASXL1 mutations reported in the literature.

TET2, which stands for ten-eleven-translocation-2, is member of a family that includes also TET1 and TET3. TET2 is located on 4q24 and contains 11 exons. The founder of the TET family, TET1, was originally identified as a fusion partner of MLL in acute myeloid leukemia with the t(10;11)(q21;q32) translocation. One known function of TET proteins is to accomplish 5-methylcytosine hydroxylation resulting in the generation of 5-hydroxymethylcytosine (Figure 1); the significance and role of this modified base is still largely unknown, but 5-hydroxymethylcytosine has been found enriched in actively transcribed genes and in the promoters of polycomb-repressed elements that are normally activated during development of mouse embryonic stem cells.3 Targeting Tet2 in mice caused a progressive expansion of hematopoietic stem and progenitor cells leading to a myeloproliferative phenotype with splenomegaly, extramedullary hematopoiesis and marked expansion of the monocytic compartment.4,5 TET2 mutations have been discovered in a wide range of myeloid malignancies,6 including classic myeloproliferative neoplasms (approximately 14%), mastocytosis, myelodysplastic syndromes, chronic myelomonocytic leukemia (CMML; 50%) and in post-myeloproliferative neoplasm or de-novo acute myeloid leukemia. Sequential analysis of the presence of TET2 mutations during the progression of myeloproliferative neoplasms has shown that these mutations may precede or follow the JAK2V617F mutation6,7 or occur at the time of disease transformation to acute myeloid leukemia.8 Mutations are scattered over the gene and consist of small insertions, deletions and nonsense mutations, all expected to result in a loss-of-function of the protein, and missense mutations affecting conserved amino acids in catalytically active regions. TET2 alterations are most commonly heterozygous, suggesting that TET2 haploinsufficiency may be a mechanism sufficient for transformation, as indicated also by the phenotype of Tet mice. Inhibition of TET2 catalytic activity is also driven by the neomorphic IDH1/2 mutant proteins.9

EZH2, located on 7q36.1, encodes for the PcG enhancer of zeste homolog 2, the catalytic component of the polycomb repressive complex 2 (PRC2) that methylates histone H3 at lysine 27 (H3K27me3). The SET domain of EZH2 (and EZH1) is specifically involved in the trimethylation of K27. H3K27me3 is a marker of inactive chromatin, as opposed to H3K4 trimethylation which is a marker of transcriptionally active status (Figure 1). EZH2 also associates with DNA-methyltransferases to direct DNA methylation. Macro- and micro-deletions of the genomic region containing EZH2 have been found in about 10% of myelodysplastic syndromes, with a few subjects presenting loss-of-heterozygosity due to acquired uniparental disomy.10,11 Mutations of EZH2 have been reported in patients with primary myelofibrosis, myelodysplastic syndromes, and myelodysplastic syndromes/myeloproliferative neoplasms;10–12 they are scattered throughout the gene and include missense, nonsense and premature stop codons resulting in loss of function. Both monoallelic and biallelic mutations have been described. Contrariwise, an activating Tyr641missense mutation has been identified in lymphomas.13 Thus, by controlling chromatin structure and gene accessibility, EZH2 may behave as a tumor suppressor or oncogene depending on the cellular context.

Figure 2.(A) Schematic representation of the human ASXL1 gene and protein. The conserved N-terminal ASX and the C-terminal plant homeodomains (PHD) are represented by light blue and blue boxes, repectively. Orange bars show HP1, LSD1 and SRC1 binding sites; red bars show nuclear receptor binding motifs (NR BOX). (B) Protein localization of described frameshift (red triangle), nonsense (black dot) and missense (yellow diamond) mutations of ASXL1 (from publications listed in Table 1). Amino acid changes that were reported as unknown possible single nucleotide polymorphism are not depicted here.

ASXL1 encodes the Additional SeX combs–Like protein-1 which is one of the three mammalian homologs of Drosophila Additional Sex Comb (Asx) gene, named after the fact that Asx deletion caused homeotic transformation due to dysregulation of Hox genes, whose spatially and quantitatively appropriate expression is essential for the anterior-posterior specification of axial structures during mammalian development. Constitutional de novo nonsense mutations of ASXL1 have recently been described in half of the subjects with Bohring-Opitz syndrome (MIM605039), a disorder characterized by severe intellectual disability, distinctive facial features and multiple congenital malformations.14 ASXL1 maps to human chromosome 20q11.21, consists of 12 exons and encodes a protein composed of 1,541 amino acids (Figure 2A). All mammalian ASXL proteins have conserved sequence features: the amino-terminal ASX homology region, which contains two of the three putative nuclear receptor box domains, and a carboxy-terminal plant homeo domain finger (Figure 2A). It is a member of the enhancer of trithorax and polycomb (ETP) family that enlists proteins required for both the maintenance of activation and silencing of gene expression by modifying chromatin configuration. For example, ASXL1 can interact with retinoic acid receptor in the presence of retinoic acid and enhance the transcription of some genes while repressing that of others, depending on the cell context.15

The fine details of the mechanism of action of ASXL1 are not well defined yet, but the protein is involved in distinct multiprotein complexes that bind to and modify chromatin at target gene regions. Scheuermann et al. demonstrated that ASXL1 exists in a complex, named polycomb repressive deubiquitinase, with BAP1, a ubiquitin carboxy-terminal hydrolase that removes monoubiquitin from histone 2A in nucleosomes.16 ASXL1 also associates with the histone acetyltransferase SRC-1, the histone methyltransferase MLL and forms a ternary complex with heterochromatin protein-1 (HP1) and the histone demethylase LSD. Thus, ASXL1 has pleiotropic and context-dependent repressive or activating effects on transcription through chemical modification of histones.

Frameshift mutations, nonsense mutations, and large 20q11 deletions of ASXL1 have been described in 10–15% of myeloproliferative neoplasms and myelodysplastic syndromes, 40% of CMML (particularly in the myeloproliferative subset, 60%), in refractory anemia with ring sideroblasts and thrombocytosis, a few patients with chronic myelogenous leukemia and 15–20% of acute leukemias (Table 1). Most ASXL1 mutations are found in exon 12, spanning the region from Tyr591 to Cys1519, and disrupt the protein downstream of the ASX homology domain with loss of the plant homo domain (Figure 2B). Germline targeted disruption of Asxl1 in mice resulted in embryonic/perinatal death while in the few mice who survived to birth only mild hematopoietic defects were detected with no evidence of myelodysplastic or myeloproliferative disorder.17 From a genetic point of view, the mutations in Asxl1 deleted mice differ from the mutations seen in patients, which usually result in the deletion of the plant homeo domain finger while sparing its N-terminal motifs; this would suggest that ASXL1 mutations generate a dominant-negative protein that can inhibit its wild-type counterpart.

In some studies, ASXL1 mutation was associated with an unfavorable outcome in acute myeloid leukemia, CMML and myelodysplastic syndromes, while there is not enough information on this aspect in classic myeloproliferative neoplasms because of the relatively small series of patients (Table 1). In this issue of the Journal, Stein et al. report on 166 patients with myeloproliferative neoplasms who were analyzed for exon 12 ASXL1 mutations.18 Extending previous results, they detected ASXL1 mutations very rarely in PV and ET, while the frequency reported in patients with myelofibrosis (included patients with primary and post-PV/post-ET myelofibrosis) was significantly higher (36%). Phenotypic correlations revealed a higher prevalence of anemia-directed therapy in ASXL1-mutated patients; however, the relatively low number of cases (n=77) examined hampered analysis of prognostic correlations.

Several lines of evidence indicate that mutations in tyrosine kinase (such as JAK2V617F and similar) are not sufficient for disease initiation and progression; rather, the JAK2V617F mutation can provide a proliferative advantage to progenitors during differentiation, allowing clonal dominance in late stages of differentiation, but it does not appear to target the stem cell.19 On the other hand, it is becoming clearer that abnormalities of proteins involved in epigenetic regulation due to mutations contributed to the pathogenesis of classic myeloproliferative neoplasms and related myeloid malignancies, possibly targeting pivotal mechanisms affecting stem cell fate.20 Furthermore, recent data suggest that there is a link between mutated tyrosine kinases and epigenetic regulators, since JAK2V617F has been shown to displace the repressor heterochromatic protein HP1α from chromatin by phosphorylating histone H3 at Tyr41,21 and affect the methylosome by disrupting the association between protein arginine methyltransferase 5 (PRMT5) and its cofactor MEP50 following uncontrolled PRMT5 phosphorylation22 (Figure 1).

Yet, there are many more questions left than answers provided. What is the hierarchy of mutations of epigenetic and tyrosine kinase genes in the events that lead to cellular transformation in myeloproliferative neoplasms? Indeed, does a defined hierarchy exist, or do mutations occur randomly? To which mutation, or group of mutated genes with overlapping functions, is the myeloproliferative neoplasm stem cell addicted? Is epigenetics a target for therapy in addition to/better than tyrosine kinase inhibition? Can known mutations predict response to treatment? These are complex questions that will certainly require even more complex experimental models. Perhaps, one piece of information that can be gained more rapidly from carefully performed clinical studies is whether any of these molecular abnormalities, singly or in combination, contribute to the phenotypic variability of myeloproliferative neoplasms and influence their prognosis.

Footnotes

• Alessandro M. Vannucchi, MD, is Associate Professor of Hematology at the University of Florence. He is active in the field of myeloproliferative neoplasms and serves as the Chairman of the AGIMM group, a network of Italian investigators whose research is focused on myeloproliferative neoplasms (http://www.progettoagimm.it) supported by the Associazione Italiana per la Ricerca sul Cancro, which contributed to this work (#10005; IG9034). Flavia Biamonte, BiSci, is currently a PhD student at the University of Florence.
• Related Original Article on page 1462
• Financial and other disclosures provided by the author using the ICMJE (www.icmje.org) Uniform Format for Disclosure of Competing Interests are available with the full text of this paper at www.haematologica.org.