The role of ASXL1 in hematopoiesis and myeloid malignancies

Abstract

Recent high-throughput genome-wide sequencing studies have identified recurrent somatic mutations in myeloid neoplasms. An epigenetic regulator, Additional sex combs-like 1 (ASXL1), is one of the most frequently mutated genes in all subtypes of myeloid malignancies. ASXL1 mutations are also frequently detected in clonal hematopoiesis, which is associated with an increased risk of mortality. Therefore, it is important to understand how ASXL1 mutations contribute to clonal expansion and myeloid transformation in hematopoietic cells. Studies using ASXL1-depleted human hematopoietic cells and Asxl1 knockout mice have shown that deletion of wild-type ASXL1 protein leads to impaired hematopoiesis and accelerates myeloid malignancies via loss of interaction with polycomb repressive complex 2 proteins. On the other hand, ASXL1 mutations in myeloid neoplasms typically occur near the last exon and result in the expression of C-terminally truncated mutant ASXL1 protein. Biological studies and biochemical analyses of this variant have shed light on its dominant-negative and gain-of-function features in myeloid transformation via a variety of epigenetic changes. Based on these results, it would be possible to establish novel promising therapeutic strategies for myeloid malignancies harboring ASXL1 mutations by blocking interactions between ASXL1 and associating epigenetic regulators. Here, we summarize the clinical implications of ASXL1 mutations, the role of wild-type ASXL1 in normal hematopoiesis, and oncogenic functions of mutant ASXL1 in myeloid neoplasms.

Introduction

Myeloid malignancies are characterized by aberrant clonal expansion and differentiation defects of hematopoietic stem cells (HSCs), hematopoietic stem progenitor cells (HSPCs) or myeloid progenitor cells. Most myeloid malignancies are associated with high mortality due to limitations of the available therapeutic agents and high relapse rate. To investigate the causative mutations of myeloid malignancies, genome-wide sequencing studies have been performed and have revealed the mutational landscape [1–4].

An epigenetic modulator, Additional sex combs-like 1 (ASXL1), is one of the most frequently mutated genes in a variety of myeloid neoplasms such as myelodysplastic syndromes (MDS) [5–7], acute myeloid leukemia (AML) [7–9], myeloproliferative neoplasms (MPN) [10–16] and chronic myelomonogenous leukemia (CMML) [14, 17–20], and its mutations are always associated with poor prognosis. Additionally, ASXL1 mutations are frequently found in clonal hematopoiesis (CH) [also called clonal hematopoiesis of indeterminate potential (CHIP)], precursor states for hematologic neoplasms with somatic mutations in the absence of diagnostic criteria for hematologic malignancies [21–23]. Therefore, understanding the mechanism by which ASXL1 mutations contribute to myeloid transformation is clinically important. To understand the functions of ASXL1, ASXL1 knockdown or Asxl1 knockout mice studies have been performed [24–26]. These studies demonstrated that ASXL1 knockdown promoted the development of MDS/MPN disease and ASXL1 depletion resulted in impaired hematopoiesis due to loss of interaction with polycomb repressive complex 2 (PRC2). On the other hand, most ASXL1 mutations exist in the last exon and would produce C-terminally truncated mutant proteins of ASXL1 (hereinafter referred as to mutant ASXL1) by escaping from nonsense-mediated-decay [8, 27]. Overexpression of mutant ASXL1 impaired myeloid differentiation and induced MDS in mouse transplantation models [28]. There is also growing evidence indicating that the physiological expression of mutant ASXL1 protein perturbs hematopoiesis and promotes myeloid transformation by altering histone modifications in both a dominant-negative and gain-of-function manner [29]. In addition, novel promising therapeutic strategies targeting ASXL1 mutated malignancies have been investigated [30–33].

In this review, we will summarize the clinical significance of ASXL1 mutations in myeloid malignancies. We will also describe recent findings of ASXL1 functions from biochemical and biological perspectives, and will then introduce potential targeted therapies for myeloid malignancies harboring ASXL1 mutations.

Members of mammalian ASXL family

Mammalian ASXL family genes (ASXL1, ASXL2 and ASXL3) are paralogs of Drosophila Additional sex combs (Asx) [34, 35]. Asx was originally identified as an enhancer of the trithorax and polycomb group (ETP) genes to regulate Hox gene expression [36, 37]. Polycomb group (PcG) genes repress [38, 39], while trithorax group (TrxG) genes activate Hox gene expression [40, 41]. Thus, Drosophila Asx is involved in both gene activation and repression. In addition, Schermann et al. revealed that Asx and Calypso, the human ortholog of BRCA1-associated protein 1 (BAP1), formed a Polycomb-repressive deubiquitinase (PR-DUB), which removes monoubiquitination of histone H2A at lysine 119 (H2AK119ub) [42]. Collectively, Drosophila Asx is now thought to integrally control gene expression through exerting a variety of epigenetic modifications.

Mammalian ASXL1 is ubiquitously expressed [43]. Human ASXL1 gene is located on chromosome 20q11 and encodes a 1541 amino acids–protein [44]. ASXL1 has an N-terminus ASXN domain, an ASX homology (ASXH) domain at the N-terminus region, and a plant homeodomain (PHD) finger at the C-terminal region (Fig. 1). ASXN, ASXH, and PHD domains are shared among all three mammalian ASXL family proteins. The ASXN domain is structurally similar to a forkhead-box domain and predicted to be essential for the DNA-binding ability of ASXL family proteins [45]. The ASXH domain is highly conserved from Drosophila to mammalian and is also called as DEUBAD (deubiquitinase adaptor) because this domain binds a deubiquitinase BAP1 [42], suggesting the importance of the interaction between BAP1 and ASXL1. The PHD domain is a histone- or DNA-binding module, and recognizes different histone modification subtypes such as unmethylated H3K4 (H3K4me0) and trimethylated H3K4 (H3K4me3) [46, 47].

Fig. 1.

Schematic representation of the structure of wild-type ASXL1 (ASXL1-WT) and C-terminally truncated mutant ASXL1 (ASXL1-MT). Their known interacting partners and post translational modifications are also shown. *Binding sites are not identified

Germline mutations of ASXL1 and ASXL3 are identified in patients with Bohring–Opitz syndrome, which is characterized by severe developmental disorders [48, 49]. ASXL2 germline mutations are associated with the Shashi-Pena syndrome, which is a neurodevelopmental syndrome [50]. ASXL1 and ASXL2 are ubiquitously expressed in a variety of tissue, whereas ASXL3 expression is restricted to lymph node, eyes, lungs, skin, brain, and pituitary gland [43].

A recent study showed that ASXL2 was essential for cardiac development and skeletal or metabolic homeostasis [51]. In myeloid malignancies, ASXL2 mutations are frequently found in AML harboring RUNX1-ETO fusion gene, whereas the frequency of ASXL2 mutations in other myeloid malignancies is much lower than that of ASXL1 mutations [52]. Interestingly, however, ASXL2 mutations are more frequently associated with RUNX1-ETO than ASXL1 mutations, making this particular fusion gene unique among many fusion genes. Asxl2-deficient mice showed more severe impaired hematopoiesis than Asxl1-deficient mice and development of MDS-like disease [53–55]. These results indicate that wild-type ASXL2 plays crucial roles as well as a tumor suppressor role in normal hematopoiesis. ASXL3 mutations are mainly detected in prostate cancers and pancreatic cancers, whereas the mutations are rarely found in hematological malignancies [56]. Although, ASXL2 and ASXL3 share conserved critical domains with ASXL1, the frequency of ASXL1 mutations are much higher than those of ASXL2 and ASXL3 mutations. The diversity of mutation frequencies within the ASXL family could be due to the differences in their unique binding partners, their binding sites on chromatin or histone modifications recognized by the PHD domain.

Clinical implications of ASXL1 mutations in myeloid malignancies

Somatic ASXL1 mutations are recurrently found in various myeloid malignancies including myelodysplastic syndromes (MDS) [5–7], acute myeloid leukemia (AML) [7–9] and myeloproliferative neoplasms (MPN) such as chronic myelogenous leukemia (CML), chronic neutrophic leukemia (CNL) and primary myelofibrosis (pMF) [10–16]. ASXL1 mutations are most frequently identified in patients with MPN/MDS overlap syndrome including chronic myelomonocytic leukemia (CMML) (50%) [14, 17–20] and juvenile myelomonocytic leukemia (JMML) [57, 58]. ASXL1 mutations are also detected in other myeloid malignancies such as blastic plasmacytoid dendritic cell neoplasm (BPDCN) [59] and systemic mastocytosis [60–62]. Additionally, ASXL1 mutations are found in aplastic anemia, a common cause of acquired bone marrow failure [63, 64]. Conversely, ASXL1 mutations are rarely found in lymphoid neoplasms [65].

The majority of ASXL1 mutations are frameshift or nonsense mutations localized at the last exon, exon 12. ASXL1 mutations frequently coexist with the following mutations; DNA methylation-related genes (TET2 [1], IDH1 [66], IDH2 [8, 66, 67]), spliceosomes (U2AF1 [68], SRSF2 [69]), transcriptional factors (CEBPA [9], RUNX1 [8, 67, 70, 71], GATA2 [72]), signal transducers (NRAS [14], JAK2 [70]), STAG2 [70] and SETBP1 [73–76]. However, ASXL1 mutations are mutually exclusive to DNMT3A [8, 67], FLT3-ITD [8, 67, 71, 77], NPM1 [8, 71, 77, 78] and SF3B1 [79] mutations. These positive and negative associations of mutations should be considered in functional analyses of these mutations.

ASXL1 mutations in acute myeloid leukemia

ASXL1 mutations are found in 5–11% of AML patients [71, 80] and independently confer poor prognosis [8, 9, 67, 71, 77]. ASXL1 mutations in AML are more common in older patients [9, 67, 71], in secondary leukemia [67] and in male patients [9, 67, 71]. In AML, ASXL1 mutations frequently coexist with RUNX1 mutations [8, 67, 71] and IDH2 mutations [67, 81], and are positively associated with FAB M0 karyotype [71, 77], t(8; 21) [52, 71, 82], trisomy 8 [67, 71] and del(7q)/− 7 chromosomal aberrations [67].

Notably, RUNX1 is the most frequently mutated gene in ASXL1-mutated AML. Coexistence of ASXL1 and RUNX1 mutations is related to poor prognosis in AML patients [67]. We previously reported that a RUNX1 frameshift mutation (RUNX1 S291fsX) indeed cooperates with an ASXL1 mutation to develop MDS/AML in a mouse model [29]. Further studies are required to reveal the precise mechanism by which ASXL1 mutation and RUNX1 mutation cooperatively induce myeloid malignancies.

ASXL1 mutations in myelodysplastic syndromes

ASXL1 mutations are found in 11–21% of patients with MDS and are also associated with adverse outcomes in MDS patients [1, 5, 83]. ASXL1 mutations are more frequently detected in patients with high-risk cases of MDS [6, 7]. DNA hypomethylating agents (HMA) such as azacitidine or decitabine are used for high-risk MDS patients. A recent study showed that TET2 mutations confer improved response to HMA; however, there was no association between ASXL1 mutations and response to HMA as there was with TET2 mutations [84]. Another study demonstrated that ASXL1 mutations are associated with shorter overall survival in MDS patients treated with HMA [85].

In MDS patients ASXL1 mutations frequently coexist with SETBP1 mutations [73–76]. SETBP1 mutations are localized in the SKI homologous region, resulting in increased stability of the SETBP1 protein [73, 76]. The presence of SETBP1 mutations is reported to be associated with quicker leukemic transformation of MDS and shorter survivals. In fact, Inoue et al. demonstrated that SETBP1 mutations rapidly drive leukemic transformation of MDS with ASXL1 mutations both in patients and in a mouse model [73].

ASXL1 mutations in chronic myelomonocytic leukemia

The ASXL1 mutation is the most frequently (40–50%) detected mutations in CMML patients. CMML patients harboring ASXL1 mutations have poorer prognosis [17, 18, 86, 87] and are categorized as a high-risk leukemic transformation group [17, 18]. Prognostic scores, including ASXL1 mutational status, divides CMML patients into three groups with distinct outcomes [17]. In CMML patients, ASXL1 mutations frequently coexist with TET2 mutations. Additional TET2 mutations are associated with shorter survival in the presence of ASXL1 mutations [88], while patients harboring TET2 mutations in the absence of ASXL1 mutations are categorized as favorable risk groups [89]. In CMML patients, hypomethylating agents are effective, but patients harboring ASXL1 mutations present a lower overall response rate (ORR) [90].

ASXL1 mutations in clonal hematopoiesis

Along with TET2 and DNMT3A mutations, ASXL1 mutations are frequently detected in clonal hematopoiesis (CH) as well [22]. Especially, CH is characterized by the presence of a somatic mutation common with hematological neoplasia without cytopenia nor dysplasia. CH is an independent risk factor in progression of myeloid malignancies [21, 23]. CH is also prevalent in aplastic anemia, and clones carrying ASXL1 mutations tend to increase in size over time [64].

A recent study revealed that CH carriers with DNMT3A, TET2, ASXL1 and JAK2 mutations are associated with atherosclerosis and coronary heart disease. Consistent with these clinical observations, Tet2-deficient mice showed enhanced progression of atherosclerosis than control mice [91, 92]. A recent study revealed that lack of Dnmt3a also accelerated atherosclerosis in mice [93]. Further studies are required to clarify whether CH with ASXL1 mutations also accelerate the development of atherosclerosis.

CH is frequently detected in solid tumor patients, particularly after chemotherapy [94]. The presence of CH in solid tumors is associated with higher recurrence ratio and adversely affects survival. It seems that chemotherapy promotes CH; PPM1D and TP53 mutations are particularly related to prior chemotherapy in CH with solid tumors [94]. Recently, there is a series of evidence that PPM1D mutations drive CH and confer resistance to chemotherapy [95, 96], but ASXL1 mutations that are unassociated with prior chemotherapy are frequently found in CH with solid tumors. On the other hand, it is also possible that CH enhances the growth of solid tumors. It will be interesting to investigate whether CH with ASXL1 mutations influence the growth of solid tumors.

The role of ASXL1 in normal hematopoiesis

To understand the roles of ASXL1 in normal hematopoiesis, several groups engineered and analyzed Asxl1 knockout mice (Table 1). Fisher et al. engineered and analyzed a constitutive Asxl1 knockout mouse. Constitutive disruption of Asxl1 led to partial perinatal lethality. Constitutive loss of Asxl1 also showed impaired B and T lymphopoiesis and impaired myeloid differentiation [97]. Wang et al. showed that heterozygous genetic Asxl1 knockout mice (Asxl1^+/−) developed MDS/MPN [26]. Asxl1 loss led to an increase in apoptotic and mitotic cells in the bone marrow. Asxl1 loss also exhibited reduced hematopoietic stem cell (HSC)/hematopoietic stem progenitor cell (HSPC) populations and impaired hematopoietic repopulation ability. In addition, Zhang et al. demonstrated that deletion of Asxl1 cooperated with Nf1 haplo-insufficiency to activate multiple oncogenic pathways such as MYC, NRAS and BRD4, promoting myeloid transformation [98].

Table 1.

Asxl1 knockout and mutant Asxl1 expressing mice studies

Mice	Peripheral blood phenotypes	HSC/HSPC phenotypes	Myeloid malignancies	Histone modifications	Gene expressions	References
Asxl1tm1Bc mutant mice	Decreased matured B-cell	Decreased formation of myeloid/erythroid colonies	Asxl1tm1Bc mutant mice did not develop AML up to 58 weeks	Not described	Not described	[97]
Mx1-Cre/Vav-Cre conditional Asxl1 knockout mice	Leukocytopenia and anemia in old (> 6 months) mouse	Increase in LT-HSCs and LSK fractions	Development of Progressive MDS-like disease	Reduced global level of H3K27me3	Increased expression of Hoxa7/9	[25]
Constitutive Asxl1 knockout mice	Leukocytopenia, anemia, thrombocytopenia	Decrease in LSK fractions, increase in GMP fractions	Asxl1+/− mice developed mild MDS-like disease	Reduced global level of H3K27me3/H3K4me3	Increased expression of Hoxa5/7/9/10	[26]
Retroviral mutant ASXL1 expression in mouse BMT	Leukocytopenia, anemia, thrombocytopenia	Not described	MDS-like disease after a long latency (> 1 year)	Reduced global level of H3K27me3	Increased expression of miR125a and Hoxa9	[28]
Asxl1 Y588X mutant expressing transgenic mice	Anemia	Increase in ST-HSCs and LSK fractions	A part of mice developed myeloid malignancies	Increased level of H3K122ac at Prdm16 promoter locus	Increased expression of Prdm16	[33]
Vav-Cre Rosa26 mutant Asxl1 knockin mice	Decreased in RBC count	Decrease in LT-HSCs and LSK fractions	Knockin mice alone did not develop MDS/AML but promoted MDS/AML development with RUNX1 mutant	Reduced global level of H2AK119ub/H3K4me3, reduced level of H3K27me3 at Hoxa loci	Decreased expression of Id3, Runx1, Sox6 and Tjp1	[29]
Constitutive locus Asxl1 G643Wfs mutant knockin mice	Leukocytosis and increase in RBC count in aged (18 months) male mice	Competitive serial transplantation assay showed disadvantage	Knockin mice alone did not develop myeloid malignancies within 18 months	Change in distribution of H3K27me3 peak	Not described	[115]
Constitutive locus Asxl1 G643Wfs mutant knockin mice	Leukocytopenia/thrombocytosis in aged (12 months) mouse	Decrease in LT-HSCs and LSK fractions	A part of mice developed MDS/MPN disease in long latency (18–24 months)	Reduced level of H2AK119ub at p16Ink4a locus	Increased expression of Hoxa7/9/10 and p16Ink4a	[116]

Abdel-Wahab et al. reported that hematopoietic cell-specific deletion of Asxl1 induced an MDS-like disease. They generated conditional Asxl1 knockout mice by crossing mice bearing floxed Asxl1 alleles with Vav-Cre or IFN-α-inducible Mx1-Cre transgenic mice [25]. Deletion of Asxl1 in hematopoietic cells resulted in age-dependent leukopenia and anemia with dysplasia. In the bone marrow of Asxl1^−/− mice, the number of HSC/HSPC was increased, but the repopulating ability of these cells were impaired. They also showed that Asxl1 and Tet2 double knockout mice developed MDS more rapidly than Asxl1^−/− or Tet2^−/− mice. Zhang et al. found that systemic deletion of Asxl1 produced more severe hematological phenotypes than conditional deletion of Asxl1, implicating an important role for Asxl1 in the microenvironment to support hematopoiesis. They further showed that bone marrow stromal cells derived from CMML patients had decreased expression of ASXL1, and that loss of Asxl1 in the bone marrow niche led to a decrease in long-term (LT)-HSCs and myeloid lineage skewing in mice [99]. In human CD34-positive cord blood cells, it was shown that ASXL1 knockdown resulted in reduced erythropoiesis and impaired erythrocyte enucleation [100].

Taken together, these studies demonstrated an essential role of wild-type ASXL1 in maintaining normal hematopoiesis. Asxl1 deletion leads to impaired progenitor differentiation and often promotes the development of myeloid malignancies.

ASXL1 interaction partners

Schermann et al. revealed that, the mammalian ASXL1, like drosophila Asx and a deubiquitinase Calypso, bound the mammalian BAP. They also showed that ASXL1 and BAP1 formed a Polycomb-repressive deubiquitinase (PR-DUB), which removes monoubiquitination of histone H2A at lysine 119, catalyzed by PRC1 complexes [42]. Wild-type ASXL1 interacts with EZH2, EED and SUZ12 as well, main components of the polycomb repressive complex (PRC) 2 to help PRC2 functions [24]. Wild-type ASXL1 protein contributes to repress their target genes such as posterior HOXA genes via collaboration with PRC2 to induce a representative histone repressive mark H3K27me3. Therefore, ASXL1 depletion results in global reduction of the trimethylation of histone H3 at lysine 27 (H3K27me3), a representative repressive mark, leading to derepression of posterior HOXA genes. It was also reported that knockdown of wild-type Asxl1 caused myeloid transformation in concert with a NRAS mutant [24]. In addition, Wang et al. revealed that lineage⁻ c-Kit⁺ cells of Asxl1-knockout bone marrow cells exhibited global reduction of both H3K27me3 and H3K4me3 [26]. Inoue et al. showed that ASXL1 interacted with OGT and HCFC1 by mass spectrometry, and found that the knockdown of ASXL1, OGT or HCFC1 decreased global levels of H3K4me3 and attenuated myeloid differentiation of HL-60 cells [31]. Previous reports showed that the OGT/HCFC1 complex bound and recruited trithorax homologues, such as MLL1, SET1/COMPASS and MLL5 [101–103]. These results indicate that wild-type ASXL1 could play pivotal roles as a scaffold to control the levels of H2AK119ub, H3K27me3 and H3K4me3, leading to epigenetic control of gene expression.

In addition, wild-type ASXL1 was shown to interact with non-histone proteins; ASXL1 directly bound AKT1 and ASXL1 deficiency led to p27-dependent cell cycle arrest, resulting in cellular senescence [104]. ASXL1 also interacts with the cohesion complex, including SMC1A, SMC3, and RAD21, and ASXL1 depletion leads to impaired telophase cohesion separation [105]. Moreover, ASXL1 interacts with RNA polymerase II (RNAPII) complex to regulate RNAPII transcriptional activity [99].

These findings demonstrated that ASXL1 interacts with a variety of molecules, important for transcription and translation, and that its loss or mutations cause aberrant histone modifications and dysregulated transcription as well as other cellular functions such as cell division and cell signaling, leading to various diseases (Fig. 2).

Fig. 2.

Overview of effects on histone modifications by wild-type ASXL1 (ASXL1-WT) and C-terminally truncated mutant ASXL1 (ASXL1-MT)

Posttranslational modifications of ASXL1

Notably, posttranslational modifications of ASXL1 influence its stability and function. Inoue et al. demonstrated that ASXL1 was ubiquitinated at lysine 351. The deubiquitinase USP7 stabilizes ASXL1 by removing polyubiquitin chain [106]. ASXL1 lysine 351 is subject to not only polyubiquitination but also monoubiquitination, in the presence of BAP1 [30]. Interestingly, monoubiquitination of mutant ASXL1 at lysine 351, in turn, activates the catalytic function of associating BAP1. Recent mechanistic analysis of mutant ASXL1 protein revealed the ‘gain of function’ features of ASXL1 mutations. BAP1, a strong interacting partner of ASXL1, is frequently mutated in renal cell carcinoma, mesothelioma and uveal melanoma, implicating BAP1 as a tumor suppressor [107–109]. However, BAP1 is rarely mutated in acute myeloid leukemia [110]. There are a series of experimental evidence that BAP1 plays tumor-promoting roles in myeloid neoplasms. Balasubramani et al. showed that the cancer-associated ASXL1 mutant protein aberrantly enhanced the catalytic function of BAP1, leading to a profound decrease in H2AK119ub [111]. Sahtoe et al. also biochemically demonstrated that the ASXH domain of ASXL1 was essential in increasing BAP1′s affinity to ubiquitin on H2A [112]. We showed the mutually reinforcing effects between the monoubiquitinated form of mutant ASXL1 and BAP1 in myeloid leukemogenesis by dysregulating HOXA and IRF8 genes [30], which are responsible for leukemogenesis and monopoiesis, respectively. We also demonstrated that depletion of endogenous BAP1 abrogated the leukemogenesis induced by mutant ASXL1, demonstrating pivotal roles of BAP1 in mutant ASXL1-induced cell transformation. Recently, Daou et al. showed that monoubiquitination of wild-type ASXL2 at lysine 370, which corresponds to lysine 351 of ASXL1, was indispensable for activation of the catalytic function of BAP1, and was catalyzed by UBE2E family proteins [113]. Whether monoubiquitination of mutant ASXL1 at lysine 351 is also catalyzed by UBE2E family proteins remains to be elucidated. In addition to ubiquitination, Inoue et al. demonstrated that glycosylation of ASXL1 at serine 199 by OGT (O-linked N-acetylglucosamine transferase) was important for its stability [31]. Functional significance of other modifications of ASXL1 such as phosphorylation, sumoylation, and methylation remains to be elucidated.

Mutant ASXL1 protein gains functions leading to myeloid transformation

As described above, Asxl1 deficiency leads to the development of myeloid diseases in mouse models, suggesting that ASXL1 mutations are loss-of-function mutations. However, accumulating evidence suggests that mutant ASXL1 proteins gain functions that promote myeloid leukemogenesis. Most ASXL1 mutations in myeloid malignancies are heterozygous frameshift or nonsense mutations localized near the 5′ end of the last exon [20]. Mutant ASXL1 transcripts are, therefore, predicted to escape from nonsense-mediated decay, resulting in production of the C-terminally truncated ASXL1 protein [114]. In cell lines derived from patients with hematological malignancies, mutant ASXL1 proteins were indeed detected by western blot and mTRAQ-based mass spectrometric analyses [27].

Hence, several groups have investigated whether the presence of the C-terminally truncated forms of ASXL1 protein induce myeloid transformation. Inoue et al. showed that mutant ASXL1 proteins (ASXL1-MT) interacted with PRC2 components and interfere with its catalytic activity. Forced expression of ASXL1-MT inhibited wild-type ASXL1 functions and caused MDS/AML development in mouse bone marrow transplantation models via derepression of miR125a and Hoxa genes caused by decreased H3K27me3 [28]. Yang et al. established C-terminally truncated mutant of Asxl1(Asxl1^Y588X)-expressing transgenic mice mimicking human ASXL1 Y591X mutation and demonstrated that transgenic Asxl1^Y588X expression led to myeloid malignancies [33]. Nagase et al. engineered a conditional Rosa26 locus ASXL1-MT knock-in mice (Asxl1-MT KI mice) mimicking human ASXL1 E635RfsX15 mutation, derived from patients with MDS/AML, and characterized the phenotype [29]. Asxl1-MT KI mice showed mild anemia and a modest block in erythroid differentiation associated with increased number of platelets, and repopulation ability of HSCs was attenuated. However, Asxl1-MT KI mice did not develop any hematological malignancies. Co-expression of a RUNX1 frameshift mutation cooperatively induced MDS/AML in Asxl1-MT KI mice. In addition, a retrovirus-mediated insertional mutagenesis study exhibited the susceptibility of Asxl1-MT KI bone marrow cells to myeloid leukemia. Thus, mutant Asxl1 promotes leukemia susceptibility.

Several groups generated and analyzed Asxl1 mutant knock-in mice at the endogenous Asxl1 locus. Hsu et al. established endogenous locus Asxl1^G643fs mutant knock-in mice mimicking human ASXL1 G646WfsX12 mutation (Asxl1^tm/+) [115]. Asxl1^tm/+ mice showed enhanced colony-forming activity of HSPCs and modestly impaired repopulation ability of HSCs. They showed that MN1 overexpression was observed in patients harboring ASXL1 mutations, and that MN1 overexpression increased the frequency of long-term culture initiation cells. However, Asxl1^G643fs mutant knock-in mice alone did not develop hematological malignancies within 18 months of follow-up. On the other hand, Uni et al. generated endogenous locus knock-in mice of Asxl1^G643fs mutant and identified different phenotypes [116], although it is not clear why theoretically the exact same KI mice gave different phenotypes. The locus KI mice developed by Uni et al. presented decreased number of HSC and increased apoptotic cells, and leukopenia and thrombocytosis were observed at 12 months old, with some mice developing MDS/MPN-like disease after a long latency period (about 18–24 months). Consistent with the previous mouse studies of mutant ASXL1, expression of Hoxa genes in Asxl1^G643fs/+ mice was dysregulated. In addition, they focused on upregulation of senescence-related genes including p16Ink4a in Asxl1^G643fs/+ mice because young Asxl1^G643fs/+ mice (3 months old) showed myeloid-skewing differentiation like aged mice. In relation to this observation, it was previously reported that the ASXL1/BAP1 axis was implicated in upregulation of p15Ink4b, supported by the fact that the promoter activity of INK4B-ARF-INK4A locus was suppressed by H2AK119ub modification [117]. Uni et al. demonstrated that wild-type, but not mutant ASXL1 proteins, interacted with BMI1, a key component of PRC1. The level of H2AK119ub was decreased at the p16Ink4a promoter locus, and Ink4a expression was derepressed in Asxl1^G643fs mutant knock-in mice. They also found that p16Ink4a knockout rescued decreased HSC numbers and aberrant apoptosis in Asxl1^G643fs mutant knock-in mice.

Collectively, these findings indicate that mutant ASXL1 at physiological expression levels alone is insufficient to induce myeloid transformation but impairs hematopoiesis and promotes susceptibility to myeloid malignancies by altering histone modifications. The distinct phenotypes of Asxl1 mutant knock-in mice among several groups could be caused by the differences in the cites of Asxl1 mutations or the levels and the hematopoietic lineages of Asxl1 expression.

Potential therapies for myeloid malignancies harboring ASXL1 mutations

Recent studies pave the way to novel therapeutic strategies for ASXL1-mutated myeloid malignancies. First, ASXL proteins/BAP1 complex promotes gene activation via opposing PRC1-mediated monoubiquitination of H2AK119 [118]. As described above, ASXL1-MT, but not wild-type ASXL1, strongly enhanced the catalytic activity of BAP1, resulting in profound reduction of H2AK119ub [30, 111]. In hematopoietic cells, hyperactive ASXL1-MT/BAP1 complex upregulates HOXA genes resulting in myeloid transformation [30]. Therefore, enzymatic activity of BAP1 or BAP1–ASXL1 binding is a potential therapeutic target for ASXL1-mutated myeloid malignancies. Guo et al. also revealed that the endogenous Bap1 activity is essential for pathogenesis of myeloid malignancies of Asxl1^Y588X transgenic mice [119].

In addition, it has been shown that ASXL1-MT, but not wildtype ASXL1, bound Bromodomain-containing 4 (BRD4) [33], a well-known oncoprotein in myeloid malignancies [120]. BRD4 activates pTEFb complex and induces acetylation of H3 at lysine 122 (H3K122Ac), resulting in phosphorylation of RNA polymerase II and gene activation. In the Asxl1^Y588X transgenic mice, the level of H3K122Ac at the promoter locus of Prdm16 was increased, resulting in dysregulated expression of Prdm16 [33]. Bone marrow cells from Asxl1^Y588X transgenic mice showed higher sensitivity to the BRD4 inhibitor than those from normal mice.

A previous study showed that combined expression of ASXL1-MT and SETBP1-MT rapidly developed MDS/AML in mice and the leukemia cells showed repression of TGFβ pathway genes [73]. Nano-liquid chromatography–mass spectrometry analysis revealed physical interaction between mutant ASXL1 and HDAC1 [30]. Saika et al. demonstrated that decrease in acetylation levels of histone H3K14 and H4K5 at TGFβ pathway genes in leukemia cells transformed by ASXL1-MT and SETBP1-MT [32]. They also showed that mutant ASXL1-induced leukemia conferred high sensitivity to an HDAC inhibitor, vorinostat. Vorinostat restored acetylation of histone H3K14 and H4K5 and the expression of TGFβ pathway genes.

On the other hand, it is effective to reactivate the functions of wild-type ASXL1 which are weakened by hemizygous ASXL1 mutations. Wild-type ASXL1/OGT complex is required for maintaining the level of H3K4me3 [31]. Depletion of ASXL1 or OGT led to impaired myeloid differentiation and global loss of the level of H3K4me3. In addition, OGT directly bound and stabilized wild-type ASXL1. Therefore, enhancing OGT activity is a reasonable strategy for restoring tumor suppressive functions of wild-type ASXL1. Intriguingly, an OGA inhibitor, which elicits the OGT activity, was effective in suppressing growth of leukemia cells expressing the mutant ASXL1 by restoring the tumor suppressor roles of wild-type ASXL1–OGT axis [31].

Taken together, inhibition of either BAP1, BRD4, HDACs or OGA has been shown to suppress leukemia with ASXL1 mutations in mouse models. These findings need to be validated using patient derived xenograft (PDX) models in future studies.

Conclusions and future perspectives

ASXL1 mutations are often associated with poor prognosis. Therefore, it is important to understand the precise mechanisms by which ASXL1 mutations contribute to myeloid transformation. Recent biological analyses demonstrated that mutant ASXL1 plays pivotal roles in leukemogenesis and leads to increased susceptibility to myeloid transformation by altering histone modifications. Meanwhile, unlike other epigenetic factors such as EZH2 and TET2, ASXL1 itself has no catalytic function. Hence, ASXL1 binding partners have been intensively investigated and biochemical analyses of these binding partners have shed light on the potential therapeutic strategies for myeloid malignancies harboring ASXL1 mutations.

While mutant ASXL1 causes dysregulations of histone modifications, resulting in myeloid malignancies, wild-type ASXL1 should also play crucial roles in epigenetic regulations under the physiological conditions via interacting a variety of epigenetic factors. In addition, ASXL1 have various post-transcriptional modifications probably induced by outside stimuli. Therefore, investigation of epigenetic control by wild-type ASXL1 may clarify how the outside stimuli are converted to the transcriptional profiles via altering epigenetics.

Compliance with ethical standards

Conflict of interest

The authors declare that they have no conflict of interest.