Introduction

Acute myeloid leukemia (AML) comprises a heterogeneous group of neoplastic disorders characterized by increased myeloid cells in peripheral blood (PB) and bone marrow (BM). Despite high remission rates after therapy, 60 to 70 % of patients with AML die within 5 years after the initial diagnosis [1]. Accumulating evidence supports the notion that AML originates from a special cell population of leukemic stem cells (LSCs) which are responsible for the proliferation of more differentiated leukemic progenitors (LPs) [2–4]. The majority of LSCs in AML patients are in a quiescent state, which makes it difficult to eradicate [5, 6]. For successful LSC-targeted therapy, the identification of differences between LSCs and hematopoietic stem cells (HSCs) is crucial.

LSCs have been reported to reside in the CD34+CD38− cell population [7]. Recently, a method using aldehyde dehydrogenase (ALDH) for the detection and separation of HSCs and LSCs from the same AML patients was developed [8–10]. ALDH is expressed at a high level by primitive stem and progenitor cells in BM. Cells with high ALDH contain residual normal HSCs and can be distinguished from CD34+CD38− LSCs with low ALDH.

Hes1 is reported to affect cell differentiation and cell cycle in various tissues including hematopoietic tissue [11]. Overexpression of Hes1 has been reported to inhibit the cycling of hematopoietic stem and progenitor cells in vitro and cell expansion in vivo, in association with upregulation of p21, a cycle inhibitor and downstream effector of Hes1 [12]. Recently, it is reported that Hes1-mediated tumor suppressive roles in MLL-AF9 AML mice, probably by downregulating FLT3 expression [13]. In our previous studies, we found that expression of Hes1 was downregulated in AML patients, and activation of Hes1 could inhibit the growth of AML cells. According to the expression of Hes1, we divided AML cases into Hes1-high group and Hes1-low group and found that Hes1 was an independent prognostic factor for AML. However, the different expression of Hes1 among HSCs, LSCs, and LPs from the same AML patients and the mechanism of regulation of Hes1 expression have not been evaluated. Here we report that miR-9 is the upstream regulator of Hes1. Furthermore, knockdown of miR-9 inhibited the proliferation of AML CD34+ cells in vitro and prolonged the survival of mice in a xenotransplant mouse model.

Materials and methods

Patient samples and cells

BM samples were obtained from adult AML patients (≧22 years of age) without any treatment at diagnosis. Patients with antecedent hematological disease or therapy-related AML were excluded. All subjects gave informed consent, and the study protocol was approved by the Ethics Committee of Tianjin Medical University Cancer Institute and Hospital. Analysis of baseline morphology, cytogenetics, molecular markers, and cell surface antigens was performed as part of the routine clinical evaluation of the patients. Blood mononuclear cells (BMNCs) were isolated by Ficoll-Hypaque (Sigma Diagnostics) separation. The clinical characteristics of the patients were shown in Table 1.

Table 1 Characteristics of AML patients included in the study
Full size table

Cell sorting

ALDH activity was assayed using ALDEFLUOR Kit according to the manufacturer’s instructions (Stem Cell Technologies, #01700). Cells were labeled with CD34-phycoerythrin (PE) and CD38-fluorescein isothiocyanate (FITC, BD Pharmingen, #555822, 555459). Cells were analyzed using a FACS Aria II cell sorter (BD Biosciences). HSCs were defined as SSClowCD34+CD38−ALDHhigh cells, LSCs as SSClowCD34+CD38−ALDHlow cells, and LPs as SSClowCD34+CD38+ cells [8–10].

Real-time PCR

Total RNA was extracted with TRIzol (Invitrogen). Reverse transcription was achieved using QuantiTect Reverse Transcription Kit (Qiagen). Real-time PCR was performed using SYBR Green Master Mix (Qiagen) and an ABI-Prism 7500 Sequence Detector (Applied Biosystems). The parameters for the thermal cycling of PCR were as follows: 15 s at 95 °C and 60 s at 60 °C, 45 cycles. The primers for Hes1 were 5′-GCAGATGACGGCTGCGCTGA-3′ (forward) and 5′-AAGCGGGTCACCTCGTTCATGC-3′ (reverse) respectively. The primers for p21 were 5′-AATCCTGGTGATGTCCGACC-3′ (forward) and 5′-TTGCAGAAGACCAATCTG-3′ (reverse). β-Actin was used as housekeeper. The primers for β-actin were 5′-ATGGAGGGGAATACAGCCC-3′ (forward) and 5′-TTCTTTGCAGCTCCTTCGTT-3′ (reverse). All values were normalized to a control gene using 2−ΔΔCt method.

MicroRNA array analysis

Total RNA was extracted from HSCs, LSCs, and LPs and evaluated in duplicate with miRCURY™ LNA expression array (Exiqon). Data were analyzed with the Bioconductor package. The Rank Prod program was used to select differentially expressed microRNAs (miRNAs) with a cutoff P value of 0.01 and an estimated false-positive rate of 0.05.

TaqMan real-time PCR

RNA was used to synthesize complementary DNAs using the Taqman microRNA Reverse Transcription Kit (Applied Biosystems, #4366596). PCR was performed using the TaqMan Universal Master MixII (Applied Biosystems, #4440048). U6 was used as control to normalize differences in total RNA levels.

Reporter gene assay

To create 3′ untranslated region (3′UTR) luciferase reporter constructs, fragments of 3′UTR from the Hes1 gene harboring the predicted miR-9 binding sites were cloned into the psiCHECK2 vector at the 3′ end of the Renilla luciferase reporter gene (Fig. 3b). The construct was sequenced and named Hes1-psiCHECK2. 293T cells were seeded at 5 × 105 cells/well in a six-well plate. One hundred nanograms of Hes1-psiCHECK2 was co-transfected into 293T cells by using Lipofectamine 2000 (Invitrogen) on the following day with miR-9 mimics or scrambled miR-9 (Applied Biosystems) at a concentration of 25 nmol/l. After 24 h, cells were lysed and luciferase activity was measured using the Bio-Glo Luciferase Assay System (Promega, #G7941). Firefly luciferase was used to normalize the Renilla luciferase. The sequence of miR-9 mimics was 5′-UCUUUGGUUAUCUAGCUGUAUGA-3′ (forward), 5′-UCAUACAGCUAGAUAACCAAAGA-3′ (reverse). The sequence of scrambled control was 5′-UCACAACCUCCUAGAAAGAGUAGA-3′ (forward), 5′-UCUACUCUUUCUAGGAGGUUGUGA-3′ (reverse).

Lentivirus-based knockdown of miR-9

For the production of miR-9 knockdown (KD) lentivirus, short hairpin RNA (shRNA) was obtained from GeneChem and cloned into the SF-LV-shRNA-EGFP vector. As a control, the scrambled hairpin plasmid obtained from GeneChem was used. Hairpin sequence of miR-9 shRNA is AUAAAGCUAGAUAACCGAAAGU. Hairpin sequence of control shRNA is TTCTCCGAACGTGTCACGT. The plasmid was co-transfected into 293T cells with pSPAX2 and pMD2.G, using Lipofectamine 2000. Virus supernatant was harvested 48 and 72 h after transfection. CD34+ BM cells from healthy donors or AML patients were sorted with immunomagnetic beads and then were incubated for 2 days in CellGro Stem Cell Growth Medium (CellGenix) supplemented with the following recombinant human proteins before transduction: interleukin 3, FLT3 ligand, stem cell factor, and thrombopoietin.

Western blot analysis

Proteins were extracted from transfected AML cells. Then protein samples (at 30 μg per lane) were analyzed by SDS-PAGE. Immunoblotting was performed using antibodies against Hes1 (Abcam), p21 (Abcam), and GAPDH (Abcam).

Colony-forming unit assay

After transduction, CD34+GFP+ cells were cultured in a 24-well plate in 0.5 ml MethoCult medium (Stem Cell Technologies) at 37 °C at a density of 104 cells/ml in triplicate. On day 10, the colonies were counted under an inverted microscope.

Cell cycle and apoptosis analysis

Transduced CD34+GFP+ cells were permeabilized and stained with Hoechst33342 (Sigma-Aldrich) followed by 1 μg/ml pyronin Y (PY, Sigma-Aldrich). The proportion of cells in G0 phase was determined by flow cytometry with quantitation of DNA and RNA. Transduced CD34+GFP+ cells were stained with bromodeoxyuridine (BrdU, BD Pharmingen) according to the manufacturer’s protocol. The proportion of cells in S phase was assessed by flow cytometry. Transduced CD34+GFP+ cells were stained with Annexin-V-PE and 7AAD (BD Pharmingen) for 20 min at room temperature. The rate of apoptosis was also assessed by flow cytometry.

Xenograft mouse model

Nonobese diabetic/severe combined immunodeficiency (NOD/SCID) mice were provided by the Animal Centre of Institute of Hematology and Blood Diseases Hospital. CD34+ AML cells from the same patient were split into two parts: one half were transduced with miR-9 KD lentivirus and the other half were transduced with control plasmid. Six- to 9-week-old mice were sub-lethally (5 Gy) irradiated and then injected by tail vein puncture with 105 CD34+GFP+ AML cells transduced with miR-9 KD or scrambled control plasmid. The AML mouse models were validated by the application of flow cytometry to detect the expression of CD45 and CD33 in BM cells. On days 14, 24, 32, 40, and 44, the proportion of CD45+ (detected with APC-H7 mouse anti-human CD45 antibody, BD Pharmingen, #560274) AML cells in PB was determined. On day 44 after transplantation, mice were killed for analysis.

Statistical analysis

Data were summarized as means ± standard deviations (SD). The significance of differences was assessed using the Student’s t test. P values less than 0.05 were considered significant. Statistical analyses were performed using Prism version 4.0 software (GraphPad).

Results

Hes1 expression differed among HSCs, LSCs, and LPs

Our previous studies showed that AML patients could be divided into two groups based on the expression of Hes1, and Hes1 downregulation was correlated with poor prognosis of AML patients [14]. In this study, we tested the expression of Hes1 in 88 adult AML patients’ BM samples and 58 of these showed downregulation of Hes1 (data not shown). Because the cell cycle status of HSCs, LSCs, and LPs were different, we hypothesized that Hes1 expression differed among these cells in the Hes1-low group. Among the 58 Hes1-low AML patients, 38 of them were CD34+ with normal karyotype. HSCs, LSCs, and LPs were sorted as shown in Fig. 1a. Real-time PCR results showed that HSCs expressed a higher level of Hes1 than either LSCs or LPs did. LPs expressed the lowest level of Hes1 among three groups (Fig. 1b). A similar pattern was detected in the expression of p21 among HSCs, LSCs, and LPs (Fig. 1c). The expression levels of Hes1 and p21 were positively correlated (r = 0.7991, P < 0.001) (Fig. 1d).

Fig. 1
figure 1

Expression of Hes1 and p21 in HSCs, LSCs, and LPs in AML patients. BM samples from 38 AML patients were obtained. a HSCs, LSCs, and LPs were sorted by flow cytometry. The expression levels of b Hes1 and c p21 were determined by real-time PCR. d Correlation between Hes1 and p21. LPs were defined as CD34+CD38+. Then the CD34+CD38– cells were divided into HSCs and LSCs based on the expression of ALDH

Full size image

MicroRNA-9 negatively regulated Hes1 expression by interacting with 3′-untranslated region of Hes1 mRNA

To explore the mechanism of Hes1 regulation, we performed microRNA (miRNA; miR) array analyses of HSCs, LSCs, and LPs (Fig. 2a). Bioinformatic analyses using PicTar and TargetScan software indicated that miR-9, miR-96, miR-130a, miR-182, miR-301a/b, miR-454, miR-507, miR-721, miR-1271, miR-4295, and miR-3666 may target Hes1. Previous reports showed that miR-9, together with miR-23, miR-124, and miR-199b, was involved in the regulation of Hes1 expression in the nervous system [15–18]. Our miRNA array analysis indicated that the expression levels of miR-9, miR-23, miR-124, and miR-199b differed among HSCs, LSCs, and LPs. TaqMan real-time PCR results confirmed that LPs expressed the highest mean level of miR-9 than either LSCs or HSCs did. LSCs expressed a higher level of miR-9 than HSCs did while miR-9 was almost undetected in HSCs (Fig. 2b). A negative correlation between the expression levels of miR-9 and Hes1 was observed (r = 0.7391, P < 0.0001) (Fig. 2c). In contrast, the expression levels of miR-23, miR-124, and miR-199b did not differ significantly among the three cell groups as determined by TaqMan PCR (Fig. 2d).

Fig. 2
figure 2

Expression of miR-9, miR-23, miR-124, and miR-199b in HSCs, LSCs, and LPs from the same AML patients. a Clustering map showing the differential expression of miRNAs in HSCs, LSCs, and LPs (n = 2). b Expression of miR-9 (Taqman real-time PCR, n = 46). c Correlation between Hes1 and miR-9. d Expression of miR-23, miR-124, and miR-199b (TaqMan real-time PCR, n = 46)

Full size image

The conserved sequences of miR-9 binding sites in the 3′ untranslated region (3′UTR) of Hes1 were shown in Fig. 3a. To evaluate the direct interaction between the 3′UTR of Hes1 and miR-9, the Hes1-psiCHECK2 construct was co-transfected into 293T cells with miR-9 mimics or a miR-9 scrambled negative control, and luciferase activity was measured 24 h later. The result showed that luciferase activity was approximately 30 % lower in miR-9-transfected cells than in cells transfected with the non-targeting scrambled miRNA (Fig. 3c).

Fig. 3
figure 3

Identification of miR-9 as a regulator of Hes1. a Conserved sequences of miR-9 binding sites in the 3′UTR of Hes1. b A schematic diagram of the constructed reporter vector. In this vector, Renilla luciferase was used as a primary reporter gene, and firefly luciferase was used to normalize transfections. c Luciferase activity in 293T cells co-transfected with miR-9 and luciferase reporter vectors. Controls were transfected with both luciferase reporter vectors and scrambled miR-9. The data were means with standard errors of the means (n = 3)

Full size image

miR-9 regulated growth of AML cells

The above results indicated that miR-9 downregulated Hes1 expression by targeting the 3′UTR of Hes1 messenger RNA (mRNA). Given that Hes1 plays an important role in cell proliferation, miR-9 may also regulate this process. To examine the functional role of miR-9 in leukemic cells, we inhibited the expression of miR-9 in AML CD34+ cells (mainly LPs and leukemia cells) by lentiviral transduction with a miR-9 KD construct containing green fluorescent protein (GFP). Expression of miR-9, Hes1, and p21 was monitored by real-time PCR and western blot showing that miR-9 was downregulated while Hes1 and p21 were upregulated after transduction (Fig. 4a, b). Analysis of colony-forming units (CFU) showed that the total colony numbers and the mean number of cells in each colony were significantly lower in CD34+miR-9 KD-GFP+ cells from healthy donors or AML patients than in control-transfected counterpart cells (Fig. 4c, d), indicating that miR-9 KD inhibited cell growth. CFU assay demonstrated that erythroid burst-forming units and granulocyte, monocyte, granulocyte/monocyte, and mixed-lineage CFU decreased when CD34+ cells from healthy donors were infected with miR-9 KD lentivirus (Fig. 4e).

Fig. 4
figure 4

miR-9 inhibited expressions of Hes1 and p21 in BMNCs of AML patients and miR-9 KD repressed AML CD34+ cell proliferation. a We detected mRNA expression levels in BMNCs after transfection with miR-9 KD lentivirus (n = 3). b Western blot results showed the protein level of Hes1 and p21 after transfection. c The colonies formed by cells transfected with miR-9 KD were smaller than the colonies formed by control cells. d Effects of miR-9 KD on colony formation of CD34+ cells from healthy donors or AML patients. e Effects of miR-9 KD on differentiation of CD34+ cells from normal donors. E erythroid burst-forming units, G granulocyte CFU, M monocyte CFU, GM granulocyte/monocyte CFU, and Mix mixed (>2 lineages) CFU. The data were means with standard deviations. *P < 0.05 (t test)

Full size image

Knockdown of miR-9 resulted in both cell cycle arrest and apoptosis in AML cells

Inhibition of cell growth can be caused by inhibition of cell proliferation and/or induction of apoptosis. To elucidate the effects of miR-9 on proliferation and cell cycle progression, cell cycle analysis was performed and revealed that after transduction of AML CD34+ cells, the proportion of S phase cells decreased (Fig. 5a) and the proportion of G0 phase cells increased (Fig. 5b). Moreover, apoptosis analysis showed that miR-9 KD induced significant apoptosis (Fig. 5c). The proportion of Q4 was 16.2 % in AML miR-9 KD-GFP+CD34+ cells from AML patients while it was 4.5 % in control-transfected counterpart cells.

Fig. 5
figure 5

Cell cycle and apoptosis analyses revealed that AML CD34+ cells transfected with miR-9 KD entered G0 phase and underwent apoptosis at a higher rate than control cells. a The proportion of S phase cells was determined with BrdU/7AAD. b The proportion of G0 phase cells was determined with Hoechst/PY. c The proportion of cells undergoing apoptosis was determined with Annexin-V/7AAD. Apoptosis% showed the percentage of Annexin V+ cells. The data in the bar graphs were means with standard deviations. *P < 0.05 (t test)

Full size image

Knockdown of miR-9 prolonged survival in AML xenotransplant mouse model

To show the therapeutic potential of targeting miR-9 in vivo, we established an AML xenotransplant mouse model by injecting 5 × 105 GFP+CD34+ cells transfected with miR-9 KD or control plasmid from the same AML patients into sub-lethally irradiated mice (Fig. 6a, n = 20 for the test and control groups each). Eight mice in the test group and ten mice in the control group developed leukemia. The results showed that survival of mice injected with AML miR-9 KD-GFP+CD34+ cells was significantly longer than that of mice injected with control cells (Fig. 6b, n = 5 in the test group and n = 7 in the control group). Leukemic cells in PB of mice injected with AML miR-9 KD-GFP+CD34+ cells appeared later and proliferated slower than that in control mice (Fig. 6c, n = 5 in the test group and n = 7 in the control group). On day 6 after transplantation, the percentage of leukemic cells in PB of control mice was 4 % on average while there was no leukemic cell in PB of miR-9 KD mice. On day 21, the average percentage of leukemic cells in PB of control mice was 82 % while it was 55 % in miR-9 KD mice. Three mice in test and control groups were killed on day 21 to analyze the BM and spleen. Leukemic cells in BM were detected by flow cytometry and spleen weights were recorded. The percentage of leukemic cells in BM of miR-9 KD AML mice was lower than that in control mice (Fig. 6d, n = 3). A low degree of splenomegaly was detected in miR-9 KD AML mice compared with control counterparts (Fig. 6e, f, n = 3).

Fig. 6
figure 6

Impact of miR-9 KD on survival of AML xenograft recipient mice. a Establishment of miR-9 KD AML mice model. b Survival in mice injected with miR-9 KD AML cells or control AML cells (n = 5 for the test group and n = 7 for the control group). c Percentage of leukemic cells in PB of control and miR-9 KD mice on days 14, 18, and 21 after transplantation (n = 5 for the test group and n = 7 for the control group). d Leukemic cells in BM of control and miR-9 KD mice on day 21 after transplantation (n = 3). e Typical spleen sizes in the two groups on day 21 after transplantation. f Spleen weight from control and miR-9 KD AML mice on day 21 after transplantation. The data in the bar graphs were means with standard deviations. *P < 0.05 (t test)

Full size image

Discussion

Current treatments for AML often fail to induce remission and are associated with high relapse rates, which may be related to resistance of LSCs [19–21]. Many researchers aim to compare the expression of genes or miRNAs between primary AML cells or cell lines and BMNCs from healthy donors. However, only a few studies compared the expression profiles among HSCs, LSCs, and LPs [22]. Here, we compared the expression levels of Hes1 and miR-9 in these types of cells from the same AML patients, taking into account of the influence of the AML microenvironment and individual differences.

However, there were limitations in our studies for that the AML samples that we used were selected. Not all AML patients showed high CD34 expression in BMNCs. We categorized AML cases into two groups based on CD34 expression [8]. One group is CD34− AML that had a very low percentage of CD34+ cells. The other one is CD34+ AML that had a relatively high percentage of CD34+ cells. In this study, the samples we used were from CD34+ AML patients.

Besides, AML is a disease with various chromosomal and/or molecular abnormalities, the prognosis of which differed greatly [23]. In addition, both tumor suppressor and oncogenic roles of miR-9 have been reported in solid tumors, while its role in leukemia is unclear. Previous studies demonstrated that miR-9 has cell context-dependent effects. It was reported that miR-9 played an oncogenic role in MLL-rearranged AML [24] while Emmrich regarded miR-9 as a tumor suppressor in pediatric AML with t(8;21) [25]. Recently, Chuang found that miR-9 was an independent poor prognostic factor for AML regardless of the karyotype and genetic mutations [26]. Considering the controversial function of miR-9 in AML with different chromosomal and/or molecular abnormalities, we used strict inclusion criteria, including adult, CD34+, and normal karyotype. All the samples went through genetic test by Taqman real-time PCR to assess mutations in NPM1, CEBPA, DNMT3A, c-Kit, IDH1, IDH2, FLT3-TKD, FLT3-ITD, WT1, and RAS.

In this study, we divided AML patients into two groups based on expression of Hes1 in primary AML cells. We confirmed that Hes1 was expressed at a high level in residual HSCs, at an intermediate level in LSCs, and at a low level in LPs in the Hes1-low group. And a positive correlation between expression of Hes1 and p21 was demonstrated. It has gradually become clear that aberrant expression of miRNAs is implicated in the pathogenesis of AML [27–30]. Interestingly, miR-9 was highly expressed in LPs in our study and was expressed at a higher level in LSCs than in HSCs, indicating a possible role of miR-9 in regulating HSC as well as LSC properties. Reporter gene assay confirmed that miR-9 regulated Hes1 expression by interacting with the 3′UTR of Hes1 mRNA. We found that inhibition of miR-9 in AML CD34+ cells led to an increase in the levels of Hes1 and p21 mRNAs compared to the level observed with a scrambled control. In vitro analysis revealed that miR-9 KD suppressed AML cell growth by cell cycle arrest and apoptosis induction. In addition, miR-9 KD reduced tumor growth in an AML xenograft mouse model. These results indicated that miR-9 regulated the proliferation of AML cells by downregulation of Hes1.

It was reported that Hes1 activation were associated with B cell lymphoma 2 (Bcl2) loss and enhanced p53/p21 expression [31]. In our studies, we found that there is a positive correlation between expression of Hes1 and p21 and activation of Hes1-induced overexpression of p21. However, the expression of Bcl2 was not tested in our study. It was reported that upregulation of antiapoptotic Bcl2 family members (i.e., Bcl2, Bcl-XL, and Mcl-1) was involved in the mediation of chemo- or radioresistance in breast and lung cancer. The Bcl2 family members have homology clustered within four conserved Bcl2 homology (BH) domains: BH1, BH2, BH3, and BH4. BH3-mimetic Bcl2 inhibitor ABT-199 binded to the hydrophobic pocket of Bcl2 with a high affinity and subsequently disrupted the antiapoptotic function of Bcl2 with potent antitumor effect in breast cancer [32]. A small-molecule Bcl2-BH4 domain antagonist, BDA-366, that binded BH4 with high affinity and selectivity may provide a strategy to improve lung cancer outcome [33]. So whether combined treatment targeting with miR-9 and Bcl2 will exhibit strong synergy against AML becomes the next study in our laboratory.

In conclusion, we showed that the expression levels of Hes1, p21, and miR-9 differed between LSCs and LPs as well as between LSCs and residual HSCs in BM cells from adult CD34+ AML patients with normal karyotype. Downregulation of miR-9 decreased the proliferation capacity of AML cells both in vitro and in vivo. Thus, by demonstrating that miR-9 suppressed Hes1, we identified that miR-9 was a potential target for the treatment of AML.