Background
Somatic mutations of additional sex comb-like protein 1 (
ASXL1) gene have been described in patients with various types of myeloid malignancies, including myelodysplastic syndromes (MDS) (15–25%), myeloproliferative neoplasms (MPN) (10–15%), 40% of chronic myelomonocytic leukemia (CMML), a few patients with chronic myeloid leukemia (CML), and 15–20% of acute myeloid leukemia (AML) [
1‐
5].
ASXL1 mutations also have been detected in myeloid blast crisis (BC) of CML patients [
5]. CMML is a clonal hematological disorder characterized by monocytosis, dysplasia, and an increased risk of progression to secondary acute myeloid leukemia (sAML) [
6,
7]. Transformation of CMML to sAML is one of the leading causes of death in CMML patients and has been associated with genetic alterations that may contribute to the leukemic transformation of CMML [
8,
9]. However, the molecular pathogenesis of the progression of CMML to sAML remains unclear.
CMML has been associated with somatic mutations in various identified genes involving epigenetic regulators, spliceosome components, transcription factors (RUNX1), and cell signaling [
6,
8,
9]. Among these, C-terminal-truncating
ASXL1 mutations (frameshift and nonsense) were associated with inferior overall survival and a high risk of AML transformation in MDS and CMML [
1,
2,
4,
10,
11]. Previous data demonstrated that ASXL1 interacts with components of the polycomb complex PRC2, namely EZH2 and SUZ12, and inhibition of ASXL1 function leading to loss of H3K27me3 histone marks [
2]. In addition to H3K27me3, recent studies have shown that ASXL1 is involved in the regulation of H2AK119 ubiquitination through interactions with BAP1 and/or BMI1 [
12,
13]. Moreover, previous data using the murine model have shown that C-terminal-truncating ASXL1 mutants inhibit myeloid differentiation and induce an MDS-like disease [
14]. Recently, Yang et al. reported that truncated ASXL1 protein functions as a gain-of-function to promote the pathogenesis of myeloid malignancies using the transgenic mouse model [
15]. We have previously found a high frequency of
RUNX1 mutations in CMML patients [
16]. We also observed that
RUNX1 and
ASXL1 mutations frequently coexisted in CMML [
17]. In addition, we found that the clonal evolution of
RUNX1 and/or
ASXL1 mutations occurred most frequently in CML with myeloid BC [
18]. We had previously shown that the biological activities of RUNX1 mutants predicted sAML transformation from CMML and MDS [
19]. Zhao et al. also found that RUNX1 mutants exhibited decreased transactivation activity as well as had a dominant-negative function on the WT-RUNX1 as a result of AML transformation in a subset of CML patients [
20].
The present study was sought to demonstrate the biological and functional evidence for a collaborative association of RUNX1 mutant and ASXL1 mutant for myeloid transformation. We identified HIF-1α targeting a new pathway which may be critical for the leukemic progression of RUNX1/ASXL1-mutated myeloid malignancies.
Materials and methods
Patient samples and cell lines
Between 1991 and 2013, 104 adult patients were consecutively diagnosed as CMML according to the WHO classification at the Chang Gung Memorial Hospital (CGMH). Mutational analyses of
ASXL1 and
RUNX1 were performed as described previously [
16,
21]. HL-60 cells were obtained from ATCC and the human leukemia cell lines, K562, THP-1, and U937, used from our stock and were authenticated by cellular morphology and STR analysis at CGMH (January–February 2017) and cultured in RPMI-1640 medium supplemented with 10% FBS, 2 mM L-glutamine, and 1× antibiotic-antimitotic in a humidified chamber with 5% CO
2 atmosphere at 37 °C. Murine myeloid leukemia 32Dcl3 (32D) cells were cultured in the presence of 1 ng/mL murine-IL-3 under similar conditions. EcoPack2-293 cell lines were cultured in DMEM medium under identical conditions.
Vector construction
The full-length cDNA of human RUNX1-WT with FLAG-tag was constructed into the NheI/NotI multiple cloning sites of lentiviral vector pCDH1-MSCV-MCS-EF1-Puro (pCMSCV, EV2) according to the standard method and verified by sequencing. Point mutant, R135T of RUNX1 gene, was generated from FLAG-RUNX1-WT using site-directed mutagenesis (KAPA HiFi HotStart, Kapa Biosystems) and confirmed by full-length DNA sequencing. ASXL1-R693X tagged with a FLAG epitope at the N-terminus was subcloned into pIRES2-EGFP-vector, then either empty vector or FLAG-ASXL1-R693X-IRES2-EGFP cassette was inserted into the pCMSCV vector to make pCMSCV-IRES2-EGFP (EV1) or pCMSCV-FLAG-ASXL1-R693X-IRES2-EGFP plasmid. Similarly, RUNX1-R135T and ASXL1-R693X were constructed into the BglII/HpaI multiple cloning sites of retroviral vector pMSCVhyg and pMSCVneo plasmids respectively. All sequences were confirmed by direct sequencing before expression in cells. The pLKO.1-puro plasmid-based shRNAs, including shLuc (luciferase shRNA, TRCN231719), human ID1-sh1 (TRCN0000019029), and ID1-sh2 (TRCN0000019030), were obtained from the National RNAi Core laboratory, Taiwan.
Lentiviral and retroviral transduction
Lentivirus production and infection were performed as our previous description [
19]. Plasmid DNAs of retroviral vector were transfected into EcoPack2-293 cells using Lipofectamine 2000. The supernatant of the transduced cells containing packaged retroviruses was collected at 48 h after transfection, centrifuged and concentrated with Retro-X Concentrator (Clontech) according to the manufacturer’s protocol. Cultured cells were spun infected in the presence of 8 μg/mL polybrene (Sigma-Aldrich) and cultured for 60 h. The infected cells were green fluorescent protein (GFP) sorted or subjected to drug selection, if necessary for 2 weeks to obtain stable clones.
Induction of differentiation and cell proliferation assays
Phorbol 12-myristate 13-acetate (PMA)-mediated myelomonocytic differentiation of U937 cells and megakaryocytic differentiation of K562 cells were induced by applying 40 nM PMA (Sigma-Aldrich) dissolved in dimethyl sulfoxide (DMSO). For control, all samples were treated DMSO only and cultured under identical conditions. To induce granulocytic differentiation of 32D cells, they were treated with 50 ng/mL granulocyte colony-stimulating factor (G-CSF) (PeproTech) for 96 h. Before assaying for differentiation and proliferation, transduced 32D, K562, or U937 cells were subjected to drug selection with the selective drug(s), if necessary. After stable transduction of 32D cells with various plasmids, cells were washed using phosphate-buffered saline (PBS) twice and cultured in complete medium with 1 ng/mL IL-3 or without IL-3 for indicated times. Cell viability was assessed by manual counting using a hemocytometer following with trypan blue staining at different time points. Similarly, the cell growth of transduced K562 cells was measured. To check cell growth of transduced K562 and U937 cells in the presence of chrysin (Santa Cruz) (HIF-1α inhibitor), cells were incubated in the presence of 30 μM chrysin for 72 h.
Flow cytometry analysis
Various differentiating inducing reagent-treated cells were collected after specific time incubation, washed in PBS, and counted. 5 × 105 cells were second time washed in PBS containing 1% bovine serum albumin then incubated with CD11b PE (eBioscience) or anti-CD61-PerCP-Ab (BD Pharmingen) or for 30 min at room temperature. Fluorescence was measured by flow cytometry (BD AriaIII) using the BD Cell-Quest Pro version 4.0.1 software.
Western blot analysis
Cell lysates were subjected to immunoblotting with the following antibodies: anti-RUNX1 (25315-1-AP) and anti-HIF-1α (20960-1-AP) obtained from ProteinTech; anti-histone-H3 (ab70550), anti-H3K4me3 (ab8580), anti-H3K27me3 (ab6002), and anti-AKT1 (phosphor S473, ab81283) from Abcam; anti-ID1 (sc-133104) and anti-ASXL1 (sc-85283) from Santa Cruz Biotechnology; anti-EZH2 (#4905) from Cell Signaling Technology; anti-AKT1+2+3 (GTX121937) from Gene Tex; and anti-flag (F3165) and anti-actin (A5441) from Sigma-Aldrich. Immunoprecipitation reaction was performed using transiently and stably expressed FLAG-RUNX1-WT/RUNX1-R135T into HEK293T and K562 cells respectively. For the endogenous interaction of ASXL1 and EZH2, we used K562 and U937 whole-cell extract. Cell lysates were subjected to immunoprecipitation with either anti-FLAG M2 affinity gel (A2220, Sigma) or Protein G Mag Sepharose Xtra (Blossom Biotenchnologies, Inc.) according to the manufacturer’s instructions. Either non-immune IgG or empty vector (EV) control was used as negative control. Cell lysate preparation and immunoblotting were performed as reported previously [
22].
Real-time quantitative PCR
Total RNA was extracted from frozen or cultured cells, patient’s bone marrow samples, and mouse spleen, liver, and BM using Trizol total RNA isolation reagent (Invitrogen). The sample was reversely transcribed to cDNA with random hexamers using the Superscript III First-Strand Synthesis System for RT-qPCR Kit (Invitrogen). The cDNA was used for quantitative PCR with iQTM SYBR® Green Supermix (BIO-Rad) according to the manufacturer’s protocol. The sequences of oligonucleotides used for quantitative PCR (qPCR) are listed in Additional file
1: Table S1. Samples were run in duplicate and transcript levels were calculated as 2(
−∆∆Ct), and the threshold cycle number for different genes was normalized to the expression of GAPDH.
Chromatin immunoprecipitation analysis
The ChIP assays were carried out according to the manufacturer’s (R&D Systems Inc. USA) protocol using Human/Mouse HIF-1α Chromatin Immunoprecipitation Kit (Cat. No. ECP1935). Cross-linked chromatin was incubated overnight at 4 °C with the anti-HIF-1α antibody on a rotating device. Immunoprecipitated DNA and input samples were cleaned up using a DNA purification kit. Purified ChIP product was quantitated by real-time qPCR using SYBR Green on an ABI Prism 7900HT Fast Real-Time PCR system. RT-qPCR quantification of ChIP was performed in duplicate using primers specific for promoter regions. ChIP was quantified relative to inputs. The following primers were used for quantitative ChIP-PCR:
ID1 (F): ACACGAACAGCAACATTATTTAGGAA, ID1 (R): GAGGCCCGAACGGAGAAG.
VEGF (F): TTGATATTCATTGATCCGGGTTT, VEGF (R): TCTTGCTACCTCTTTCCTCTTTCTG.
GAPDH (F): CTTGACTCCCTAGTGTCCTGCT, GAPDH (R): CCTACTTTCTCCCCGCTTTTT.
Gene expression microarray analysis
Gene expression analysis was carried out using Affymetrix Human Gene U133 Plus 2. Total RNA was extracted from stably transduced K562 cells using the Trizol reagent method. Amplification and biotin labeling of fragmented cDNA was carried out using the standard Affymetrix protocol. Labeled probes were hybridized to the Affymetrix GeneChip Hybridization Oven 645 and GeneChip Fluidics Station 450 and scanned. Expression data were extracted from image files produced on GeneChip Scanner 3000 7G. The scanned images were analyzed with the Standard Affymetrix protocol. GeneChip analysis data normalized with RMA by Affymetrix Expression Consol Ver 1.4 (EC 1.4) software and fold change were calculated compared to the empty vector control. The upregulated genes were selected using the criteria of undergoing a > 2.0-fold change in gene expression. The gene expression microarray data have been deposited in the Gene Expression Omnibus (GEO) database with accession number GSE99640.
Mice and BMT experiment
All animal experiments were approved by the Department of Animal Experimentation at CGMH. C57BL/6 mice (NARlabs, Taiwan) were used for BMT experiments. Mouse BMT was performed as described previously [
23]. Briefly, BM cells were isolated from 8- to 12-week-old mice which were pretreated with 5-fluorouracil (150 mg/kg) 4 days before the operation. BM cells were cultured with RPMI containing 20% FBS, 2 mM L-glutamine, 1× antibiotic-antimitotic, 100 ng/mL mouse stem cell factor, and 10 ng/mL mouse IL-3. The primary murine BM cells then were transduced with retroviruses by spin inoculation in the presence of 8 μg/mL polybrene. The infection was repeated after 48 h of the first infection, and transduced BM cells (1 × 10
6 cells/mouse) were transplanted into intraperitoneally injected busulfan (a single dose of 30 mg/kg before 3 days) mice via the tail vein [
24,
25]. Mice were monitored every day, and moribund mice were euthanized according to the animal house guideline.
Colony-forming assays were performed according to the manufacturer’s instructions (MethoCult M3434; StemCell Technologies, Vancouver, BC, Canada). Briefly, 2 × 104 transduced mouse BM cells were mixed with 2 mL MethoCult medium in duplicate cultures in 6-well tissue-cultured plate. Colonies of more than 50 cells were scored on day 8 of cultures. Serial replating assays were performed on day 8 of the previous culture. All cells were harvested, washed twice with RPMI medium, and counted. A similar number of cells were then replated, and the process was repeated for four times to check colony formation and self-renewal activity. Results from colony-forming unit (CFU) assays to assess granulocyte (CFU-G; colorless, more dense, smaller and round cells), macrophage (CFU-M; colorless, less dense, large and elongated cells), granulocyte with macrophage (CFU-GM; colorless, heterogeneous population of small, round cells and large, elongated cells) and erythroid (CFU-E; red color either very small colonies or BFU-E; clusters containing group of tiny cells in irregular shape).
Luciferase reporter assay
K562 cells were co-transfected with luciferase reporter plasmid of HRE promoter, pGL4.42[Luc-2p/HRE] (Promega, RE4001) combined with EV, RUNX1-WT, and RUNX1-R135T plasmids, and 32D cells with EV, ASXL1-R693X, RUNX1-R135T, and ASXL1-R693X + RUNX1-R135T; pEGFP-C1 (Promega) was used as internal control using TransIT-2020 transfection reagent (Mirus, Blossom Biotechnologies Inc.). Forty-eight hours later, cells were incubated with 100 μM CoCl2 (Alfa Aesar, B22031) for an additional 24 h. After 72 h, cells were harvested, lysed with passive lysis buffer (Promega), and mixed with Bright-Glo Luciferase assay reagent (Promega) for detection of luciferase activities and GFP fluorescent intensity. The relative activity of each sample was derived by normalization of each luciferase intensity with its GFP intensity and then divided by the value of EV.
Statistical analysis
The Kaplan–Meier analysis was used to evaluate survival. Differences in survival were assessed using the log-rank test. All in vitro data represented here were mean ± SD as indicated. In all analyses, P values were two-tailed, and values < 0.05 were considered significant for all analyses.
Discussion
The prognosis of CMML is poor, and effective therapies are limited; CMML remains to be a challenging hematological malignancy concerning the pathogenesis and treatment [
33]. With the introduction of next-generation sequencing technology in the past decade, it is well known that CMML is characterized by the presence of various somatic mutations of driver genes [
6,
7,
34].
ASXL1 mutations were common in CMML and frequently associated with a combination of various gene mutations [
4,
34,
35]. We had analyzed the mutational status of various mutations related to myeloid neoplasms in a cohort of 110 patients with CMML.
ASXL1 mutations were present in 29 (26.4%) of CMML patients.
RUNX1 mutations were detected in 34 (30.9%) patients ([
16] and updated). Among them, both
ASXL1 and
RUNX1 were mutated in 10% (11/110) of our patients. Moreover, 7 of the CMML patients carrying
ASXL1 mutations had sAML transformation later, 4 of them coexisted with
RUNX1 mutations at initial diagnosis of CMML, and additional 2 patients acquired
RUNX1 mutations at sAML transformation [
36]. We had observed that 8 of 11 patients harbored
RUNX1 mutations located in the runt homology domain in
ASXL1-mutated CMML [
17]. On the other hand, we could not detect
ASXL1 or
RUNX1 mutations in CML patients at initial diagnosis of chronic phase; however, acquisition of
RUNX1 and/or
ASXL1 mutations occurred in 25.5% (13/51) of patients during AML transformation. Moreover, 3 of the 6 CML-myeloid BC patients carrying
ASXL1 mutations had co-existence of
RUNX1 and
ASXL1 mutations [
18]. To further investigate the role of RUNX1 mutations to the myeloid transformation in ASXL1-mutated leukemia, we performed in vitro and in vivo expressing either
ASXL1 or
RUNX1 single mutant or combination of
ASXL1 with
RUNX1 mutant. In accordance with the augmentation of cell growth and impairment of differentiation of 32D cells, a murine myeloid cell line, self-renewal activity of primary murine BM cells, was also increased in the presence of the co-expression of ASXL1-R693X and RUNX1-R135T. Myeloid progenitor 32D-expressing
RUNX1 mutants blocked granulocytic differentiation by G-CSF [
20]. It was also reported that either ASXL1-MT alone or the cooperation of SETBP1-D868N and ASXL1-MT impaired G-CSF-induced differentiation of 32D cells [
14]. Recently, Nagase et al. reported that the expression of mutant Asxl1 resulted in the dysfunction of hematopoietic stem and progenitor cells, perturbed erythroid-lineage differentiation, and promoted leukemia transformation in vivo using conditional knock-in mouse [
37]. Previously, Goldfarb found that loss-of-function or haploinsufficiency of
RUNX1 gene impaired megakaryopoiesis [
38]. We also observed that the overexpression of
RUNX1-R135T in K562 cells, a heterozygous
ASXL1 mutant CML cell line, augmented cell growth and impaired PMA-induced megakaryocytic differentiation as evidenced by the reduction of CD61 expression with immature cell morphology in
RUNX1-R135T-expressing K562 cells. In contrast, either BCR-ABL1 or ABL1 expressions were not changed in R135T-expressing K562 cells. The K562 cell line is a multipotent leukemia cell line that undergoes megakaryocytic differentiation upon PMA induction with enhanced expression of surface antigens such as CD41 and CD61 [
26]. We also found that overexpression of
RUNX1-R135T in
ASXL1 mutant K562 cells upregulated
HOXA genes; however, the expression of H3K27me3 was not changed in mutant cells.
It has been demonstrated that overexpression of
ID1 immortalized myeloid progenitors in vitro and led to MPD in vivo [
39]. Moreover, high
ID1 expression was associated with poor outcome in AML with shorter event-free and overall survival [
40]. We found that
ID1 mRNA and protein levels were elevated in the BM, spleen, and liver samples of mice carrying both
ASXL1-R693X and
RUNX1-R135T mutants compared to either
ASXL1-R693X or
RUNX1-R135T mutant and control mice. Similarly,
ID1 expression in BM cells from CMML patients carrying both
ASXL1 and
RUNX1 mutations was not only higher than normal BM cells but also much higher than either
ASXL1-MT or
RUNX1-MT alone. ID1 plays a critical role in the leukemogenesis of AML through regulation and interaction with AKT1 [
41]. The activation of AKT signaling is an important mechanism of transformation to AML, and the effects of ID1 on leukemogenesis through AKT has been reported [
41,
42]. In line with this, we observed the upregulation of AKT1 signaling with the enhancement of ID1 expression in coexisted
ASXL1 and
RUNX1 mutant cells that would contribute to the leukemogenesis in a subset of patients with CMML or CML myeloid BC (Fig.
8).
RUNX1 and HIF-1α directly interacted with each other, in which the runt homology domain of RUNX1 was mainly involved [
43]. Overexpression of RUNX1-WT inhibited DNA binding and transcriptional activity of HIF-1 protein with reduced expression of HIF-1-targeted genes [
43]. HIF-1α is a master transcriptional regulator that maintains HSC cell cycle regulation and activates the transcription of genes that are involved in critical aspects of cancer biology, including angiogenesis, cell survival, and invasion [
43‐
45]. We observed that both RUNX1-WT and RUNX1-R135T could interact with HIF-1α; however, RUNX-R135T might have more interaction with HIF-1α. Moreover, we showed that overexpression of RUNX1-R135T or combined expression of ASXL1-R693X and RUNX1-R135T enhanced transcriptional activity of HIF-1α and its target gene,
ID1 expression. One possible explanation of this finding was that more stable RUNX1-R135T protein might compete with RUNX1-WT protein for DNA binding and β-heterodimerization and reduced RUNX1-mediated transcriptional activity in the cells. Hence, mutant RUNX1-R135T increased HIF1-α activity and its target gene expression. Previously, we systematically analyzed the biologic activities of RUNX1 mutants identified from patients with CMML and MDS by in vitro functional assays [
19]. We observed that most RUNX1 mutants had reduced abilities in DNA binding, CBF-β heterodimerization, and C-FMS gene induction, especially missense mutations at runt homology domain, but we did not use RUNX1-R135T mutant for these functional experiments in that study. HIF-1α is a well-established transcriptional regulator of VEGF. Other investigators demonstrated that Id-1 induced angiogenesis through HIF-1α-mediated VEGF activation in human endothelial cells, breast cancer, and hepatocellular carcinoma [
46‐
48]. Our results of reduction of ID1 expression by the inhibition of HIF-1α in transformed leukemia cells supports that ID1 is controlled by HIF-1α, which might be deregulated by either
ASXL1 or
RUNX1 mutation or coexisted mutant of both genes. Our ChIP data following co-expression of
ASXL1 and
RUNX1 mutations recognized the deregulation of
ID1 in myeloid leukemia cells, suggesting the role of myeloid leukemia transformation by combined mutations were at least partly attributed to the upregulated
ID1 expression (Fig.
8). Data presented here showed that either
ASXL1-R693X or
RUNX1-R135T mutant alone did not have much effect on leukemia cells; however, cooperative mutations led to the enhancement of HIF-1α recruitments to the promoter region of the
ID1 gene. This suggested that the increase was not only due to the upregulation and stabilization of HIF-1α by the RUNX1 mutant, but also ASXL1 plays a significant role in this finding. Both ASXL1 and ID1 were physically interacted with AKT1 [
41,
49], and ASXL1/AKT1/ID1 axis may regulate HIF-1α expression in combined ASXL1 and RUNX1-mutated cells of which underlying mechanism remains to be explored. Notably, the leukemic cells from our patients harboring coexisted mutations of
ASXL1 and
RUNX1 correlated with the upregulation of
ID1 gene expression, supporting the role of cooperative mutation of
ASXL1 and
RUNX1 on ID1 expression and myeloid leukemia transformation.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.