Epigenetic dysregulation of hematopoietic stem cells and preleukemic state
- 1.8k Downloads
Recent genetic analyses have revealed that premalignant somatic mutations in hematopoietic cells are common in older people without an evidence of hematologic malignancies, leading to clonal hematopoietic expansion. This phenomenon has been termed clonal hematopoiesis of indeterminate potential (CHIP). Frequency of such clonal somatic mutations increases with age: in 5–10% of people older than 70 years and around 20% of people older than 90 years. The most commonly mutated genes found in individuals with CHIP were epigenetic regulators, including DNA methyltransferase 3A (DNMT3A), Ten–eleven-translocation 2 (TET2), and Additional sex combs-like 1 (ASXL1), which are also recurrently mutated in myeloid malignancies. Recent functional studies have uncovered pleiotropic effect of mutations in DNMT3A, TET2, and ASXL1 in hematopoietic stem cell regulation and leukemic transformation. Of note, CHIP is associated with an increased risk of hematologic malignancy and all-cause mortality, albeit the annual risk of leukemic transformation was relatively low (0.5–1%). These findings suggest that clonal hematopoiesis per se may not be sufficient to engender preleukemic state. Further studies are required to decipher the exact mechanism by which preleukemic stem cells originate and transform into a full-blown leukemic state.
KeywordsHematopoietic stem cells Clonal hematopoiesis CHIP Preleukemic state Epigenetic regulators
Recent studies using next-generation sequencing analysis have uncovered premalignant genetic changes in healthy individuals with no evidence of hematologic malignancies. This phenomenon has been termed clonal hematopoiesis of indeterminate potential (CHIP), which is associated with increased risk of developing hematologic malignancy and decreased overall survival [1, 2]. The most frequently mutated genes found in individuals with CHIP were epigenetic modifiers, including DNA methyltransferase 3A (DNMT3A), Ten–eleven-translocation 2 (TET2), and Additional sex combs-like 1 (ASXL1). Notably, these epigenetic modifiers are also recurrently mutated in various myeloid malignancies, suggesting an essential role of these mutations in early phase of leukemia development. Indeed, functional correlation between epigenetic modifiers and normal/malignant hematopoiesis has been a focus of intense study for almost a decade. These studies have shown that mutations in epigenetic modifiers exert pleiotropic effect in leukemogenesis, including hematopoietic stem cell (HSC) self-renewal/differentiation, transcriptional regulation of oncogenes/tumor suppressors, and DNA damage response [3, 4]. In this review, we will focus on genetic basis and clinical features of CHIP, roles of mutations in epigenetic regulators in preleukemic stem cells (Pre-LSC) and myeloid transformation, and hypothetical model of leukemogenesis.
Clonal hematopoiesis of indeterminate potential and age-related clonal hematopoiesis
Early analysis of inactivation patterns of X-linked genes such as glucose-6-phosphate dehydrogenase (G6PD) or the androgen receptor (AR/HUMARA) revealed the clonal origin of myeloid malignancies [5, 6]. Although similar techniques have shown that age-related clonal skewing of hematopoietic cells is relatively common after the age of 55–65 years, until recently the general notion was that this clonal hematopoiesis is rarely associated with hematologic malignancy [7, 8]. Recent study has identified TET2 mutations in 5.6% of older women with age-related X-inactivation skewing clonal hematopoiesis without evidence of hematologic malignancy, suggesting that initial driver mutations can drive age-related clonal hematopoiesis . In addition, DNMT3A mutations, but not leukemic blast-specific NPM1 mutations, were identified in non-neoplastic normal T cells derived from acute myeloid leukemia (AML) patients, further providing evidence that mutations in some epigenetic modifiers can serve as premalignant genetic changes .
Clonal hematopoiesis and preleukemic state
The above studies have demonstrated that the presence of CHIP is a strong risk factor of developing hematologic malignancy [12, 13]. Along this line, individuals with clonal mutations in the Swedish cohort showed an increased risk of hematologic cancer diagnosis (hazard ratio (HR) 12.9) and death (HR 1.4) compared to age-matched individuals without mutations . In addition, study in population-based cohorts revealed that the presence of CHIP is associated with an increase in the risk of hematologic malignancy (HR 11.1) and all-cause mortality (HR 1.4) compared to age-matched controls . Importantly, however, annual risk of leukemic transformation in individuals with CHIP was only 0.5–1%. Moreover, death from hematologic malignancy alone could not explain the observed increase in overall mortality. Rather, the most significant cause of decreased overall survival associated with CHIP was an increased propensity for smoking, thrombosis, and cardiovascular diseases, including coronary artery disease and stroke [12, 13]. Indeed, atherosclerosis-prone low-density lipoprotein receptor-deficient (Ldlr−/−) mice reconstituted with Tet2-deficient BM cells led to clonal hematopoietic expansion and a marked increase in atherosclerotic plaque in vivo due to increased NLRP3 inflammasome-mediated interleukin-1β secretion in Tet2-deficient macrophages, providing functional relevance of somatic TET2 mutations in hematopoietic cells in atherosclerosis development . These data indicate that, although clonal genetic changes or cooperating mutations may lead to alteration of normal hematopoiesis, clonal hematopoiesis per se may not be sufficient to engender preleukemic state, which evolves with high frequency into full-blown leukemia.
Clonal somatic mutations in aplastic anemia
Clonal hematopoiesis is also identified in non-malignant hematologic diseases. Acquired aplastic anemia (AA) is characterized by peripheral blood cytopenia with hypocellular bone marrow. However, given the resemblance of clinical presentation and the lack of clear diagnostic criteria distinguishing AA and hypoplastic myelodysplastic syndromes (MDS), the exact clinical diagnosis is often challenging . Previous studies have shown the evidence of clonal hematopoiesis in AA. Indeed, around 50% of patients with acquired AA have expanded populations of paroxysmal nocturnal hemoglobinuria (PNH) clones . Moreover, transient cytogenetic abnormalities have been reported in acquired AA patients with no apparent evidence of MDS . Despite these observations, until recently, the significance and molecular basis of clonal hematopoiesis in AA have been elusive.
Recent targeted deep sequencing study analyzed 150 acquired AA patients with no morphological evidence of MDS and detected clonal somatic mutations in 19% of these cases . The most frequently mutated genes were ASXL1, DNMT3A, and BCL6 corepressor (BCOR) (Fig. 1), with the highest median mutant allele burden of 31% in ASXL1 mutations . In this study, patients with clonal somatic mutations had longer disease duration and shorter telomere lengths compared to patients without mutations. In addition, AA patients with clonal somatic mutations with disease duration of longer than 6 months were associated with 40% risk of transformation to MDS . More recently, targeted deep sequencing of myeloid cancer candidate genes was performed in 439 patients with AA and identified mutations in these genes in approximately one-third of AA patients . Consistent with previous study, the most commonly mutated genes were DNMT3A, ASXL1, BCOR/BCORL1, and phosphatidylinositol glycan anchor biosynthesis class A (PIGA) (Fig. 1), most of which showed less than 10% variant allele frequencies [17, 18]. Of note, DNMT3A-mutated and ASXL1-mutated clones tended to increase in size over time, whereas the size of BCOR/BCORL1-mutated and PIGA-mutated clones was more likely to decrease or remain stable . Furthermore, mutations in BCOR/BCORL1 and PIGA correlated with a better response to immunosuppressive therapy (IST) and favorable survival, whereas mutations in a subset of genes including DNMT3A and ASXL1 were associated with a poorer response to IST, inferior overall survival, and progression to MDS or AML . Together, these studies suggest that a significant proportion of AA patients with no morphological evidence of MDS possess mutations in myeloid cancer-related genes including epigenetic regulators, indicating functional relevance of epigenetic modifiers in clonal hematopoiesis in AA. In the following chapter, we will review pleiotropic effect of major epigenetic modifiers in HSC regulation and leukemogenesis, mutations of which were reported in individuals with clonal hematopoiesis.
Roles of epigenetic modifiers in clonal hematopoiesis and leukemogenesis
Around 50% of DNMT3A mutations in AML are heterozygous missense mutation in Arg882 (R882, most commonly R882H), which is located in the catalytic domain of the enzyme [20, 21]. In physiologic condition, DNMT3A functions as a tetramer, comprising either two homodimers or heterodimers with DNMT3L, a DNMT family member that lacks a methyltransferase catalytic domain . Heterodimerization of DNMT3A with DNMT3L enhances its methyltransferase activity [23, 24]. Although mouse Dnmt3a R878H (corresponding to human R882H) mutant protein can still interact with wild-type Dnmt3a and Dnmt3b, co-expression of wild-type and mutant form in murine embryonic stem (ES) cells led to inhibition of the wild-type DNA methylation ability, consistent with dominant-negative effect of DNMT3A R882H mutations . Moreover, DNMT3A R882H mutant was shown to inhibit wild-type de novo methylation activity by disrupting the formation of its functional tetramers , further confirming the dominant-negative role of this mutant.
Several studies have reported the functional relevance of DNMT3A in hematopoiesis. Initial study demonstrated that Dnmt3a/Dnmt3b double-knockout HSCs, but not single deficient HSCs, show disrupted HSC self-renewal . More extensive analysis in vivo using hematopoietic tissue-specific conditional Dnmt3a-deleted animals revealed progressive expansion of long-term HSC compartment with impaired differentiation in Dnmt3a-null cells by incomplete epigenetic repression of HSC-specific genes . More recently, hematopoietic tissue-specific conditional loss of both Dnmt3a and Dnmt3b resulted in enhanced HSC self-renewal and a more severe differentiation block than Dnmt3a single-null cells, partly due to activated β-catenin signaling . In agreement with the murine data, recent targeted deep sequencing study has identified recurrent DNMT3A mutations at high allele frequency in highly purified HSC population as well as in normal T-cell compartment in AML patients . In addition, DNMT3A mutant HSCs showed a multilineage repopulation advantage over non-mutated HSCs in xenografts, suggesting a fundamental role of DNMT3A mutation in establishing Pre-LSCs . Given DNMT3A is also one of the most recurrently mutated genes in individuals with CHIP and AA patients with clonal hematopoiesis [11, 12, 13, 17, 18], these evidences indicate that DNMT3A mutation is clearly responsible for clonal hematopoietic expansion.
Further studies focused on the function of DNMT3A mutation in leukemogenesis. In vivo analysis using a retroviral transduction of DNMT3A R882H mutant construct and bone marrow transplantation (BMT) assay showed development of chronic myelomonocytic leukemia (CMML)-like disease in mutant recipients, possibly through increased CDK1 protein level and enhanced cell-cycle activity . Similarly, mice transplanted with Dnmt3a-null whole bone marrow (BM) or HSCs developed a spectrum of hematologic malignancies, including MDS, AML, myeloproliferative neoplasms (MPN), and T- and B-cell acute lymphocytic leukemia (ALL) [31, 32]. Furthermore, primary mice with conditional hematopoietic Dnmt3a loss caused fully penetrant MPN with myelodysplasia (MDS/MPN) in vivo, with cell-autonomous aberrant tissue tropism and marked extramedullary hematopoiesis with liver involvement . These studies underscore a significant relevance of DNMT3A as a tumor suppressor in vivo. Of note, recent study demonstrated that DNMT3A R882-mutated AML cells show chemoresistance to anthracyclines by impaired nucleosome eviction and chromatic remodeling in response to anthracyclines, which resulted from attenuated recruitment of histone chaperone SPT-16, leading to defective DNA damage response . This result, at least partially, explains why DNMT3A R882-mutated AML patients show an inferior outcome when treated with standard-dose daunorubicin-based induction chemotherapy [35, 36] and how DNMT3A R882 mutant cells persist and drive relapse after induction therapy .
The TET family of proteins was first reported with the cloning of TET1 as a fusion partner of MLL1 in patients with t(10;11)(q22;q23) AML . TET proteins (TET1-3) are Fe(II)- and α-ketoglutarate- (α-KG)-dependent mammalian DNA oxidases that catalyze the conversion of 5-methylcytosine (5mC) to 5-hydroxymethylcytosine (5hmC) (Fig. 2) . The discovery of this new modification on DNA methylcytosine has provided a novel insight into DNA demethylation pathways. The TET enzymes can also oxidize 5hmC to 5-formylcytosine (5fC) and 5-carboxylcytosine (5caC) . Furthermore, 5caC can be directly recognized and repaired by thymine DNA glycosylase (TDG)-mediated base-excision repair (BER) to generate unmethylated cytosine, leading to active DNA demethylation (Fig. 2) . Alternative active DNA demethylation through the activation-induced cytidine deaminase (AID)-APOBEC DNA repair pathway has also been reported. The first step of this pathway is the conversion of 5hmC to 5-hydroxymethyluracil (5hmU) by AID/APOBEC, followed by TDG- or single-strand-selective monofunctional uracil DNA glycosylase (SMUG1)-mediated BER to generate unmethylated cytosine (Fig. 2) [41, 42]. Of note, Aid-deficient BM cells demonstrated a significant accumulation of 5hmC compared to wild-type cells, consistent with deaminase activity of Aid on 5hmC . On the other hand, 5hmC may also lead to passive DNA demethylation, as DNMT1, a maintenance methyltransferase that methylates unmethylated cytosine in the daughter strand upon DNA replication, cannot recognize 5hmC . Although these studies suggest a critical role of TET enzymes in active/passive DNA demethylation, the exact biochemical mechanism of demethylation process, the exact genetic loci that are affected by TET-mediated demethylation, biological significance of each chemically modified cytosine, and their functional relevance in cancer are still under active investigation. Recent studies have also uncovered a novel role of TET proteins in chromatin modifications. Several groups have reported that TET proteins interact with O-linked β-N-acetylglucosamine (O-GlcNAc) transferase (OGT), tethering OGT to the target gene promoters and regulating gene transcription through histone H2B O-GlcNAcylation or H3K4 trimethylation [45, 46, 47]. These data indicate that TET also modifies chromatin landscape as well as DNA methylation, thereby regulating gene transcription.
TET2 is one of the major epigenetic modifiers recurrently mutated in individuals with CHIP [11, 12, 13]. Previous studies using high-throughput genome-wide sequencing have identified somatic deletions and loss-of-function mutations in the TET2 gene in 10–20% of patients with MDS/MPN [48, 49], in 10–20% of patients with AML and in 40–50% of patients with CMML [50, 51]. Additionally, TET2 mutations are also reported in lymphoid malignancies, especially at high frequency in angioimmunoblastic T-cell lymphoma [52, 53, 54, 55], suggesting a common key role of TET2 as a tumor suppressor in hematologic malignancies. Large series of clinical correlative study has demonstrated that TET2 mutations are associated with poorer prognosis in patients with intermediate-risk, cytogenetically normal AML . Although several studies have shown a decrease in global 5hmC levels in patients with TET2-mutated myeloid malignancies, changes in global levels of 5mC in these patients are not conclusive [56, 57]. While transcriptional silencing through DNA methylation at promoter CpG islands has been well characterized, the effect of TET2-mediated demethylation reactions on methylation change at CpG islands is controversial. Recent bisulfite pyrosequencing study detected hypermethylation at non-CpG island promoters in some gene loci but not at promoter CpG islands in TET2-mutated CMML patient samples . In addition, depletion of Tet2 in AML1-ETO-induced murine pre-leukemic hematopoietic cells had little impact on the methylation change at CpG islands and promoters but rather showed hypermethylation at enhancers .
A number of studies have explored the functional role of TET2 in hematopoiesis and leukemogenesis. Tet2-silenced murine BM hematopoietic stem/progenitor cells (HSPCs) showed preferential differentiation toward myeloid lineage in vitro . Consistent with murine data, TET2-silencing in human cord blood (CB) CD34+ cells led to skewed differentiation toward CD14+ monocytic lineage in ex vivo culture experiment . Indeed, Aid-deficient mice as well as AID-silenced human BM CD34+ cells also demonstrated myeloid skewing, suggesting that TET2/AID active DNA demethylation pathway might play an essential role in myeloid fate commitment . In addition, oncometabolite 2-hydroxyglutarate (2-HG) produced by Isocitrate dehydrogenase 1 and 2 (IDH1/2) mutations, recurrently seen in AML patients, inhibits TET2 function as well as other α-KG-dependent oxygenases (Fig. 2) [57, 61]. It has been shown that Wilms tumor 1 (WT1) recruits TET2 to regulate its target gene expression  and WT1 mutations lead to reduced TET2 function and decreased 5-hmC levels . These studies revealed that IDH1/2 mutations and WT1 mutations share, at least partially, common epigenetic pathogenesis in AML as TET2 mutations through altered DNA hydroxymethylation. Several groups have generated mouse models of germline or conditional Tet2 loss and reported common features in these mice, including disrupted hematopoietic differentiation, expansion of HSPC (LSK, Lin− Sca-1+ c-Kit+) compartment, and enhanced HSC self-renewal [52, 64, 65, 66, 67, 68]. Furthermore, some Tet2-null mice eventually developed myeloid malignancies in vivo [52, 64, 65]. Of note, Tet2-disrupted fetal liver common myeloid progenitors (FL-CMPs) exhibited enhanced replating capacity in vitro, implicating that Tet2 loss may potentially transform more differentiated myeloid progenitors . Additionally, microRNA-22 (miR-22), which targets TET2, was reported to be upregulated in MDS patient samples . Mice conditionally expressing miR-22 displayed increased HSC self-renewal with defective hematopoietic differentiation and developed MDS similar to Tet2-deficient mice, confirming the functional relevance of miR-22/TET2 regulatory network in myeloid transformation . Together, these data clearly show that Tet2 regulates both myeloid differentiation and clonal hematopoietic expansion, thereby functioning as a tumor suppressor.
Given that most of the individuals with CHIP alone do not progress to hematologic malignancy [12, 13] and long latency is required for developing myeloid disease in Tet2-deficient mice [52, 64, 65], TET2 loss itself is insufficient to cause hematopoietic transformation. In fact, previous studies have uncovered co-occurrence of other disease alleles with TET2 mutations in myeloid malignancies [35, 71, 72], underscoring a critical role of convergent cooperative effects of these alleles. Recent study has demonstrated cooperative function of Tet2 loss and Flt3 ITD mutations in AML development in vivo through combinatorial epigenetic remodeling of specific genetic loci . In agreement with co-occurrence of TET2 and EZH2 mutations in MDS and MDS/MPN patients, concurrent depletion of Tet2 and Ezh2 in mice developed MDS or MDS/MPN by derepression of oncogenic polycomb targets . In addition, combined Tet2 loss with AML1-ETO leads to fully penetrant AML in vivo, partially due to hypermethylation of enhancer regions thereby silencing tumor suppressors [59, 75]. Furthermore, studies from two independent groups have shown that combination of Tet2 loss and Jak2V617F resulted in aggressive MPN phenotype through both clonal HSC dominance and expansion of downstream precursor populations [76, 77]. Notably, TET2 mutations also co-occur with changes in other TET members or DNMT3A mutations in human acute B-lymphocytic leukemia and T-cell lymphoma [78, 79]. Concordant with this, Tet1/2 double-knockout mice developed lethal B-cell malignancies, and Tet2 loss combined with Dnmt3a mutation caused T-cell lymphoma/leukemia in vivo, possibly owing to dysregulated Bcl6/Myc and Notch pathway, respectively [78, 80, 81]. Collectively, these data suggest that functional convergent cooperativity of TET2 mutations and co-occurring disease alleles drives hematopoietic transformation.
ASXL1 is the human homolog of Drosophila Additional sex combs (Asx). Recently, it has been demonstrated that Drosophila Asx forms a complex with the chromatin deubiquitinase, Calypso, to form the Polycomb-repressive deubiquitinase (PR-DUB) complex, which removes monoubiquitin from histone H2AK119 thereby repressing Hox gene expression . The mammalian homolog of Calypso, BRCA-1-associated protein 1 (BAP1), also associates with ASXL1, and the mammalian BAP1–ASXL1 complex was shown to harbor deubiquitinase activity in vitro .
Analogous to DNMT3A, ASXL1 is recurrently mutated in individuals with CHIP and AA patients with clonal hematopoiesis [11, 12, 13, 17, 18]. Multiple studies have also identified mutations in ASXL1 in 15–20% of MDS, 25% of AML, and 40–60% of MDS/MPN patients [83, 84, 85]. Of note, ASXL1 mutations seem to be enriched in AML secondary to a preexisting MPN rather than in de novo AML , suggesting a crucial role of ASXL1 in myeloid transformation. Indeed, several clinical studies found that ASXL1 mutations are associated with adverse outcome in AML and MDS [35, 87, 88].
The nuclear deubiquitinase enzyme BAP1 utilizes ASXL1 as an essential cofactor. Recent genetic studies have identified recurrent BAP1 mutations in various cancers, including mesothelioma, renal cell carcinoma, and metastatic uveal melanoma [95, 96, 97]. Hematopoietic-specific conditional loss of Bap1 resulted in retention of monoubiquitinated H2AK119 and decreased expression levels of transcriptional regulator host cell factor-1 (HCF1) and OGT, leading to MDS/MPN-like disease in vivo . However, in contrast to Asxl1 loss, Bap1 deletion was shown to increase Ezh2 and H3K27me3 levels and enhance repression of PRC2 targets, sensitizing BAP1-deficient mesothelioma to pharmacologic EZH2 inhibition . These data indicate that ASXL1 and BAP1 loss may function in independent manner in myeloid transformation.
This work was supported in part by Sumitomo Life Welfare and Culture Foundation Foreign Medical Research Grant, Astellas Foundation for Research on Metabolic Disorders Foreign Medical Research Grant, and Clinical Scholars Biomedical Research Training Program Fellowship from Charles A. Dana Foundation.
- 43.Kunimoto H, McKenney AS, Meydan C, et al. Aid is a key regulator of myeloid/erythroid differentiation and DNA methylation in hematopoietic stem/progenitor cells. Blood. 2017 (in press).Google Scholar