A novel miR-375-HOXB3-CDCA3/DNMT3B regulatory circuitry contributes to leukemogenesis in acute myeloid leukemia

Human leukemic cell lines including THP1 and HL-60 (ATCC, Manassas, VA, USA) were purchased for the present study. Leukemic cell lines were cultured in RPMI 1640 supplemented with 10% fetal bovine serum (Invitrogen, Carlsbad, CA, USA) and 1% penicillin-streptomycin in humidified 37 °C incubator with 5% CO₂. Bone marrow mononuclear cells from AML patients and mononuclear cells from umbilical cord blood (UCB) were isolated by Ficoll density gradient centrifugation (GE Healthcare, Uppsala, Sweden) and cultured in StemSpan SFEM (StemCell Technologies, Vancouver, Canada), which was supplemented with human SCF, Flt3 ligand, erythropoietin, IL-3, and IL-6 at final concentrations of 10 ng/ml. UCB samples were received by the normal donor repository in the Laboratory of Internal Medicine at the First Affiliated Hospital of Wenzhou Medical University. Normal CD34⁺ cells from normal controls (NC) were isolated and enriched by immunomagnetic positive selection kit (Stem Cell Technologies, Vancouver, Canada). Cell viability was determined by the trypan-blue exclusion assay. All procedures involving human participants were in accordance with the ethical standards of the Ethics Committee of the First Affiliated Hospital of Wenzhou Medical University and the Declaration of Helsinki. All the patients gave informed consent. 5′-azacytidine (AZA, Sigma-Aldrich, St Louis, MO, USA) was dissolved in distilled water and kept at − 20 °C until used.

A total of 102 patients (Additional file 1: Table S1, Age ≤ 60) between October 2012 and October 2015 with de novo AML at the First Affiliated Hospital of Wenzhou Medical University were included. Patients were diagnosed and classified according to the French–American–British (FAB) criteria. This study was approved by the institutional ethics committee and a written informed consent in accordance with the recommendations of the Declaration of Helsinki. All AML patients (excluding M3 subtype) received first-line treatment with standard induction chemotherapy, which consisted of daunorubicin 60–90 mg/m² or idarubicin 12 mg/m² on days 1–3 and cytarabine 100–200 mg/m² on days 1–7. One or two courses of induction chemotherapy were given to patients to get complete remission (CR). AML patients were treated with 3–4 cycles of high-dose cytarabine in the consolidation therapy. For the patients who were refractory or relapsed, salvage therapy was taken. Cytogenetic analyses of the samples were performed at diagnosis through G-banding. The criteria used to describe a cytogenetic clone and description of karyotype followed the recommendations of the International System for Human Cytogenetic Nomenclature [19]. Fusion genes such as AML1-ETO were measured by qRT-PCR [20] and gene mutations such as WT1 and c-Kit were performed on all of the samples by direct sequencing as previously reported [3]. Bone marrow mononuclear cells from these 102 AML patients were isolated and the expression of miR-375 was detected. The median expression of miR-375 was used as the cutoff. Survival probabilities were estimated by the Kaplan-Meier method and differences in survival distributions were compared using the log-rank test. Overall survival (OS) was defined as being from the date of diagnosis to death or last follow-up, and disease-free survival (DFS) was defined as the time in complete remission (CR) to relapse or death due to progressive disease.

Total RNA was extracted by the miRNeasy extraction kit (Qiagen, Valencia, CA, USA) and was used as a template to synthesize cDNA for quantitative RT-PCR (qRT-PCR) analysis in a 7500 real-time PCR system (Applied Biosystems, Carlsbad, CA, USA). Mature miR-375, miR-126, and U6 snRNA were reversely transcribed using Stem-loop RT primer with miScriptIIRT Kit (Qiagen, Valencia, CA, USA). Pre-miR-375 was measured by miScript Precursor Assays (Qiagen). qPCR with SYBR Green dye (Qiagen) was used to determine the expression of mRNA and miRNA. GAPDH and U6 snRNA were used as endogenous controls for qRT-PCR of mRNA and miRNA, respectively. The relative expression levels were evaluated using the 2^-ΔΔCt method. Each sample was run in triplicate. All the primer sequences were shown in Additional file 2: Table S2.

Western blot analysis was performed using standard techniques. Briefly, transiently transfected leukemia cells were harvested and lysed by RIPA buffer (Thermo Scientific, Waltham, MA, USA). Proteins (40 μg) from the lysate were fractionated by electrophoresis through polyacrylamide gels (Bio-Rad, Richmond, CA, USA) and transferred to PVDF membranes. Blots were incubated with primary antibodies over night at 4 °C and incubated with second HRP-conjugated antibody for 1 h. Signals were measured by chemiluminescence reagents (Bio-Rad) with imaging system (Bio-Rad). The following antibody was used: HOXB3 (ab82945, Abcam, Cambridge, MA, USA); CDCA3 (ab166902, Abcam); DNMT3B (ab176166, Abcam). As necessary, blots were stripped and reprobed with β-actin antibody (Santa Cruz Biotechnology, Santa Cruz, CA, US) as an endogenous control.

About 2×10⁶ leukemic cells were washed and resuspended in PBS solution containing 0.04 mg/ml propidium iodide (Invitrogen) and 100 mg/ml RNase. The samples were analyzed by flow cytometry (Becton Dickinson, Mountain View, CA) using the CELLQuest program (Becton Dickinson) after incubation at room temperature for 30 min. Modifit software (Verity Software House, Topsham, ME, USA) was used to generate histogram to determine number of cells in each phase of the cell cycle.

293 T cells were seeded in 24-well plates at a density of 1.0×10⁵/ml per well, followed by growth for 24 h before transfection. Each well was transiently cotransfected with 100 ng pMIR-REPORT plasmid bearing the wide-type 3’UTR of HOXB3 or the 3’UTR with mutated miR-375 binding site of HOXB3, 60 pmol scramble (negative control), or miR-375 mimics (GenePharma, Shanghai, China) as well as 10 ng endogenous control vector pRL-SV40 (Promega, Madison, WI, USA) using Hiperfect transfection reagent (Qiagen). Firefly and Renilla luciferase activities were measured by the Dual-Luciferase Reporter Assay System (Promega) in cell lysates harvested after transfection for 24 h. The value of relative luciferase activity indicates the Firefly luciferase activity normalized to that of Renilla luciferase for each assay. Each experiment was performed in triplicate and repeated three times.

Genomic DNA from leukemic cells was extracted using DNA Purification Kit (QIAGEN). For the treatment with sodium bisulfite, DNA (1 mg) was modified by EpiTect Bisulfite Kit (QIAGEN). For bisulfite-sequencing analysis, upstream flanking sequence from − 260 bp − + 136 bp in the pre-miR-375 gene was amplified by PCR using the sodium bisulfite-treated DNA as template. PCR products were subcloned into pUC18 vector and five constructed pUC18 vectors were used to sequence for each sample. The methylation frequencies were calculated as methylated C/ (methylated C + unmethylated C) in every clone. For methylation-specific PCR analysis, the methylation-specific and unmethylation-specific primers were designed by MethPrimer software [21]. Primer sequences are shown in Additional file 2: Table S2.

Before transfection, leukemic cells were seeded in 6-well plates at a density of 2.0 × 10⁵/ml per well. 100 pmol special miR-375 inhibitor Agomir (GenePharma, Shanghai, China) or negative control was transiently transfected into leukemic cell lines by Hiperfect transfection reagent (Qiagen). After transfection for 48 h, cells were collected and proteins were extracted for western blot.

The paired primers based on the primary sequence of hsa-miR-375 and its flanking regions were cloned into murine stem cell virus vector pMSCV-puro (Clontech, Palo Alto, CA, USA) to construct the plasmid that expresses miR-375. To produce plasmids expressing HOXB3 and DNMT3B, human HOXB3 and DNMT3B coding sequences were amplified by PCR and then cloned into lentivirus vector pLVX-IRES-ZsGreen1 (Clontech) and pMSCV-puro (Clontech), respectively. To construct pMIR-HOXB3 3′UTR plasmid, human HOXB3 3′UTR was amplified by PCR and cloned into pMIR-REPORT vector (Ambion, Dallas, TX, USA). The mutation on miR-375-binding sites in human HOXB3 3′UTR was generated by the site-directed mutagenesis kit (Stratagene). All the primer sequences were shown in Additional file 2: Table S2. All of these plasmids were confirmed by sequencing.

HEK293 T cells (2×10⁶) were plated in 10 cm dish. After 24 h, all constructed plasmids together with negative control vectors were co-transfected with packaging plasmids into HEK293 T cells. Virus was harvested from the supernatant at 48 h after transfection and further filtered through a 0.45 μm low protein-binding-polysulfonic filter (Millipore, Billerica, MA, USA). Leukemic cells (2×10⁵/ml) were suspended in viral supernatant with 8 μg/ml polybrene (Sigma-Aldrich, St. Louis, MO, USA) and centrifuged at 2000 rpm for 2 h. Puromycin (2 μg/ml, Medchemexpress, Princeton, NJ, USA) was added into the supernatant for 1 week to select positive clones.

Gene-specific short hairpin RNAs (shRNAs) for HOXB3, DNMT3B, and CDCA3 were designed and cloned into pSIREN-RetroQ (Clontech) retroviral vectors. Control shRNA is a nonfunctional construct provided from Clontech. The sequences of shRNAs were shown in Additional file 2: Table S2. These vectors were co-transfected with packaging plasmids (Gap-pol and VSV-G) into 293 T cells to produce retrovirus. Supernatants containing retrovirus were collected at 48 h after transfection and were utilized to transduce leukemic cells.

Bone marrow mononuclear cells (blasts% > 70%) from six AML patients were harvested. CD34⁺ cells were enriched by immunomagnetic positive selection kit (Stem Cell Technologies), followed by the retrovirus transduction of MSCV-miR-375 or MSCV-NC. Retrovirus-transduced CD34⁺ cells (5×10³) were seeded on a 35-mm dish with 1.0 ml of methylcellulose medium (MethoCult™ H4434 Classic, Stem Cell Technologies). HL-60 and THP1 cells, which were transduced with MSCV-miR-375 or MSCV-NC, were seeded in the same methylcellulose medium (Stem Cell Technologies). Normal CD34⁺ cells were isolated and enriched from UCB, followed by the retrovirus transduction of MSCV-miR-375 or MSCV-NC and then were plated in the same methylcellulose medium (Stem Cell Technologies). Colonies were scored 10–14 days after plating according to the manufacturer’s protocol.

Primitive hematopoietic progenitor cells were harvested and enriched by the Mouse Lineage Cell Depletion Kit (Stem Cell Technologies) from a cohort of 6-week-old C57 mice on the sixth day after 5-fluorouracil treatment. An aliquot of enriched hematopoietic progenitor cells was added to retroviral supernatant together with polybrene in six-well-plate, which were centrifuged at 2000×g for 2 h. The media was replaced with fresh media and incubated for 24 h at 37 °C. Next day, the same procedure was repeated once. Then, an equivalent of 2.0 × 10³ cells was plated onto a 35-mm dish with 1.0 ml of methylcellulose medium (MethoCult™ GF M3434, Stem Cell Technologies). The colonies were replated every 7 days under the same condition. The colony-forming/replating assays were repeated three times.

The binding activity of DNMT3B in pre-miR-375 promoter was examined by Chromatin Immunoprecipitation (ChIP) Assay Kit (Merck-Millipore, Billerica, MA, USA) according to the manufacturer’s instruction. Briefly, treated and untreated leukemic cells were cross-linked with 1% formaldehyde for 10 min. Chromatin from nuclear extracts was sonicated to generate 200–1000 bp DNA fragments. DNA-protein complexes were immunoprecipitated with 5 μg of specific anti-DNMT3B (ab176166, Abcam). DNA was purified after the DNA-protein cross-link was reversed by heating at 65 °C for 4 h. Standard PCR reactions were performed by primer pairs (Additional file 2: Table S2). The immunoprecipitated DNA was analyzed by qRT-PCR and the amount of precipitated DNA was calculated as the percentage of the input sample.

Male athymic nude mice were purchased from SLAC (Shanghai SLAC Laboratory Animal, Shanghai, China). The animals were maintained under specific pathogen-free (SPF) conditions in the Animal Facility of Wenzhou Medical University. Animal-related experiments were performed according to the Guide for the Care and Use of Laboratory Animals (National Institutes of Health Publications) and approved by the committee for human treatment of animals at Wenzhou Medical University Institutional Guidelines. Six-week-old mice (N = 12) were randomly divided into two equal groups. A total of 1×10⁷ viable HL-60-MSCV-NC or HL-60-MSCV-miR-375 cells were resuspended with 200 μl sterile 1× PBS and injected subcutaneously into the right flank of each nude mouse. Both the long and short diameters of xenografts were measured using vernier calipers every 3 days after 2 weeks. When the experiments were terminated at 6 weeks after tumor cell inoculation, all mice were sacrificed and the tumors from each mouse were harvested and weighed. Tumor volumes were measured using the eq. V (mm³) = AхB²/2, where A is the largest diameter and B is the perpendicular diameter. Tumor lysates were prepared for western blot.

THP1 cells were transduced with lentivirus vector pLVX-IRES-ZsGreen1 (Clontech), followed by sorting to obtain GFP-positive cells. GFP-positive THP1 cells were transduced with MSCV-miR-375 or MSCV-NC, followed by puromycin selection for 1 week. Busulfan (30 mg/kg; B2635; Sigma) was intraperitoneally given to eight-week-old NSG mice 1 day before xenotransplantation. 4×10⁵ THP1-GFP-miR-375 cells or THP1-GFP-NC cells were intravenously injected into NSG mice (N = 10). When the mice became moribund, they were humanly sacrificed and survival was evaluated from the first day of the experiment until death.

To determine whether miR-375 is decreased in leukemic cells as reported in solid tumors [10, 22], we measured the levels of miR-375 in 7 leukemic cell lines, 102 primary AML blasts, and 20 CD34⁺ hematopoietic stem and progenitor cells (HSPC) from normal controls (NC). MiR-375 expression was about 60% lower in leukemic cell lines than in NC (P < 0.01) and about 76% lower in primary AML blasts than in NC (P < 0.01, Fig. 1a), respectively. We further analyzed the expression of miR-375 according to FAB subtype. No significant differences of miR-375 were found between different FAB subtypes from M1–M5 (Fig. 1b). To confirm that miR-375 is virtually expressed in HSPC, the level of a well-described stem cell gene miR-126 [23] was measured in 20 CD34⁺ cells from NC. The average expression of miR-126 is 1.34-fold higher than miR-375 in CD34⁺ cells (Additional file 3: Figure S1A, P = 0.0696). Because miR-126 is well expressed in HSPC and no difference was observed between miR-375 and miR-126 expressions, we speculate that miR-375 is virtually expressed in HSPC.

The low expression of miR-375 in AML cells prompted us to determine whether miR-375 expression is associated with clinical outcome in our adult AML cohort. AML patients were divided into two different groups according to the median miR-375 level. MiR-375 expression above or below the median was considered as higher or lower expression. As shown in Fig. 1c and d, AML patients with higher expression of miR-375 predicted better overall survival (OS, HR = 2.311; 95% CI =1.226–4.357; P = 0.007) and disease-free survival (DFS, HR = 1.997; 95% CI = 1.115–3.577; P = 0.015) compared with those with lower miR-375 expression, respectively. Furthermore, multivariate analysis indicated that the prognostic impact of miR-375 expression on OS and DFS was independent of age, sex, karyotype, and gene mutations including FLT3, CEBPA, and NPM (Additional file 4: Table S3).

Recently, emerging data have indicated that gene silencing is frequently caused by promoter hypermethylation in multiple types of cancers [24, 25]. Several typical CpG islands were found around the region encoding pre-miR-375 by MethPrimer software [21], suggesting that low expression of miR-375 might be due to DNA hypermethylation (Additional file 3: Figure S1B). To determine whether CpG islands of pre-miR-375 were hypermethylated in a tumor-specific manner, the methylation status of miR-375 was analyzed by methylation-specific PCR (MSP) in 7 leukemic cell lines, 40 primary AML blasts, and 20 NC (CD34⁺). Hypermethylation of miR-375 was observed in all 7 leukemic cell lines (Fig. 2a) and in 35 of 40 (87.5%) primary AML blasts (Fig. 2b). In contrast, a complete absence of miR-375 methylation was found in 20 NC (Fig. 2c). We further confirm that these CpG islands were densely methylated by bisulfite sequencing. A CpG map of the sequenced region comprising -260 bp − + 136 bp in the pre-miR-375 gene was indicated in Fig. 2d. CpG islands were densely methylated in HL-60 and THP1 cells but not in one NC (NC#1) by bisulfite sequencing (Fig. 2e, Left). The frequencies of miR-375 methylation in HL-60 and THP1 cells were about 82% and 65%, respectively, compared with 14% in NC#1 (Fig. 2e, Right). To explore whether DNA hypermethylation leads to the silencing of miR-375, the expression of miR-375 was measured in HL-60 and THP1 cells treated with DNA methyltransferase inhibitor 5′-azacytidine (AZA). AZA increased the expression of miR-375 in HL-60 and THP1 cells (Fig. 2f).

To determine potential miR-375-targeted genes, the microRNA.​org software (http://​www.​microRNA. org) was utilized to predict possible target genes. HOXB3 was finally chosen for the next analysis because HOXB3 is overexpressed in AML patients and promotes proliferation and self-renewal of mouse and human HSPCs [26, 27]. As reported, miRNAs regulate gene expression mainly through binding 3’-UTR of target genes [5]. MiR-375 potentially binds 3’-UTR of HOXB3 (Fig. 3a), which includes putative miR-375-binding sites and was subcloned into pMIR vectors to construct wide-type vector pMIR-HOXB3’UTR and mutation vector pMIR-HOXB3’UTR (Mut). These constructs were cotransfected into 293 T cells with either miR-375 mimics or scramble. Luciferase activity was measured at 48 h after each transfection. Cells cotransfected with miR-375 mimics exhibited approximately 60% decrease of luciferase activity in comparison with those cotransfected with scramble. However, the decrease of luciferase by miR-375 was almost abolished by the mutation in miR-375-binding sites (Fig. 3b).

To explore whether miR-375 can reduce endogenous HOXB3 expression, HL-60 and THP1 cells were transduced with retrovirus vector MSCV-NC or MSCV-miR-375. To avoid supraphysiological expression of miR-375 in leukemic cell lines, we diluted the virus titer to transduce leukemic cells and selected clones with the lowest overexpression of miR-375. The expression of miR-375 is about 20–30-fold higher in leukemic cells transduced with MSCV-miR-375 than MSCV-NC (Fig. 3c). To further confirm that miR-375 is actually overexpressed in leukemic cells transduced with MSCV-miR-375, the expression of pre-miR-375 was measured. The levels of pre-miR-375 were elevated by about 30-fold in leukemic cells transduced with MSCV-miR-375 than MSCV-NC (Additional file 3: Figure S1C). Accordingly, overexpression of miR-375 decreased both the protein and mRNA expressions of HOXB3 than NC (Fig. 3d and e). Furthermore, three primary AML blasts were transduced with MSCV-miR-375 or MSCV-NC. Similarly, overexpression of miR-375 reduced the expressions of HOXB3 in all three primary AML blasts (Fig. 3f).

We then determined whether inhibition of miR-375 modulates the expression of HOXB3. Leukemic cell lines were transfected with special miR-375 inhibitor. Inhibition of miR-375 modestly increased the protein expression of HOXB3 in HL-60 and THP1 cells (Fig. 3g). Next, we asked whether HOXB3 is upregulated in AML cells. The mRNA expressions of HOXB3 were detected in the same 7 leukemic cell lines, 102 primary AML blasts, and 20 NC. The expression of HOXB3 was 3.0-fold higher in leukemic cell lines than in NC (P < 0.01, Fig. 3h) and 3.7-fold higher in primary AML blasts than in NC (P < 0.01, Fig. 3h). Finally, we determined whether miR-375 expression is inversely associated with HOXB3. When the relative expression of miR-375 was plotted against with HOXB3, a moderate inverse correlation between miR-375 and HOXB3 was found in 102 patients with AML (R = − 0.249 and P = 0.012, Fig. 3i).

To determine whether miR-375 possesses anti-leukemia activity, proliferation and apoptosis were detected in HL-60 and THP1 cells transduced with MSCV-miR-375 or MSCV-NC. MiR-375 reduced the cell growth in leukemic cells (Fig. 4a) but did not induce cell apoptosis (data not shown). To further examine whether miR-375 affects cell cycle in leukemia cells, distribution of G0/G1, S, and G2/M phase was analyzed in HL-60 and THP1 cells transduced with MSCV-miR-375 or MSCV-NC. Overexpression of miR-375 significantly extended G0/G1 phase in HL-60 and THP1 cells (Additional file 5: Figure S2A and B), suggesting that the extended G0/G1 phase by miR-375 might inhibit proliferation of leukemia cells. Next, miR-375 notably decreased the number of colonies in leukemic cell lines (Fig. 4b) and reduced colony formation in 5 of 6 CD34⁺ leukemia stem/progenitor cells from AML patients (Fig. 4c). To further investigate the biological function of miR-375 in leukemia stem/progenitor cells, we performed a colony-forming/replating assay. Normal mouse bone marrow progenitor cells were transduced with MSCV-GFP-MLL-AF9 or MSCV-GFP-MLL-AF9 plus MSCV-miR-375 and then were plated onto methylcellulose medium, because MLL-AF9 can rapidly transform normal hematopoietic stem cells into leukemia stem cells [28]. The colonies were replated every 7 days under the same condition for another two times. Forced expression of miR-375 significantly inhibited the colony-forming capacity induced by MLL-AF9 after second and third replating (Fig. 4d). However, miR-375 did not affect the colony-forming capacity in normal CD34⁺ HSPC from umbilical cord blood (UCB, Fig. 4e). To explore whether HOXB3 plays an important role in the anti-leukemia activity by miR-375, leukemic cells were transduced with pLVX-HOXB3 or pLVX-NC. As shown in Fig. 4f, HOXB3 was significantly increased in LVX-HOXB3-transduced leukemic cells than LVX-NC-transduced cells. Furthermore, overexpression of HOXB3 partially prevented miR-375-induced cell growth arrest (Fig. 4g) and partially blocked the reduction of colony formation by miR-375 in HL-60 and THP1 cells (Fig. 4h).

As reported, HOXB3 facilitates cell proliferation through transcriptional activation of cell division cycle associated 3 (CDCA3) [17], which prevents the arrest of cell cycle progression at the G1 phase [29]. To determine whether HOXB3 regulates the expression of CDCA3 in leukemic cells, the protein expressions of HOXB3 and CDCA3 were measured in HL-60 and THP1 cells transduced with special shRNA for HOXB3 or negative control. HOXB3 expression was significantly decreased by sh-HOXB3 followed by the downregulation of CDCA3 (Fig. 5a). Having shown that miR-375 negatively regulated HOXB3 expression and inhibited proliferation and colony formation, we then determined the role of HOXB3 knockdown in the anti-leukemia activity. Knockdown of HOXB3 by special shRNA decreased the proliferation (Fig. 5b) and colony-formation ability (Fig. 5c) in HL-60 and THP1 cells compared with negative control. To exclude the possible unwanted off-targets by shRNA, another shRNA for HOXB3 (sh-HOXB3#2) were constructed. As shown in Additional file 6: Figure S3A, sh-HOXB3#2 significantly reduced the expression of HOXB3 followed by the downregulation of CDCA3. Accordingly, knockdown of HOXB3 by sh-HOXB3#2 decreased the proliferation than negative control in HL-60 and THP1 cells (Additional file 6: Figure S3B). Next, we determined whether knockdown of CDCA3 presents anti-leukemia activity. HL-60 and THP1 cells were transduced with special shRNA for CDCA3 or negative control. The protein expression of CDCA3 was significantly decreased in sh-CDCA3-transduced cells compared with negative control (Fig. 5d). Accordingly, the decrease of CDCA3 inhibited cell growth (Fig. 5e) and reduced colony formation (Fig. 5f).

HOXB3 increases the expression of DNA methyltransferase (DNMT3B), which is in turn recruited to the RASSF1A promoter, resulting in hypermethylation and silencing of RASSF1A expression in lung adenocarcinoma [18]. Therefore, we determined whether HOXB3 increases the expression of DNMT3B and further enhances DNA hypermethylation of miR-375. HL-60 and THP1 cells were transduced with special shRNA for HOXB3 or negative control. Reduction of HOXB3 by shRNA downregulated the expression of DNMT3B (Fig. 6a). Meanwhile, the forced overexpression of HOXB3 increased the level of DNMT3B (Fig. 6b), suggesting that HOXB3 regulates the expression of DNMT3B in leukemic cells. To explore whether DNMT3B facilitates the silencing of miR-375, DNMT3B was knocked down by special shRNA. Knockdown of DNMT3B (Fig. 6c) increased the expression of miR-375 (Fig. 6d), suggesting that DNMT3B possibly binds pre-miR-375 promoter and regulates DNA hypermethylation. To exclude the possible unwanted off-targets by shRNA, another shRNA for DNMT3B (sh-DNMT3B#2) was constructed. Accordingly, sh-DNMT3B#2 reduced the expression of DNMT3B (Additional file 6: Figure S3C) and increased the expression of miR-375 (Additional file 6: Figure S3D). To further confirm that DNMT3B binds pre-miR-375 promoter, two pairs of primer surrounding the putative promoter of pre-miR-375 for ChIP were designed to detect the possible binding activity of DNMT3B (Additional file 7: Figure S4A). Immunoprecipitated DNA from HL-60 and THP1 cells was analyzed by qRT-PCR. Knockdown of DNMT3B by special shRNA substantially attenuated the binding activity of DNMT3B in pre-miR-375 promoter (Fig. 6e). To determine whether DNMT3B actually methylates these regions, bisulfite sequencing was performed in HL-60 and THP1 cells, in which DNMT3B was knocked down or overexpressed. Knockdown of DNMT3B drastically decreased the methylation frequency (Fig. 6f), while overexpression of DNMT3B increased the methylation frequency (Fig. 6g). Western blot indicated successful overexpression of DNMT3B (Fig. 6h).

Finally, we determined whether HOXB3 modulates the binding activity of DNMT3B in pre-miR-375 promoter and regulates the expression of miR-375, in turn. ChIP assay was performed using anti-DNMT3B antibody in HL-60 and THP1 cells, which were transduced with sh-NC or sh-HOXB3. Knockdown of HOXB3 decreased the binding ability of DNMT3B in pre-miR-375 promoter (Additional file 7: Figure S4B) and subsequently increased the expression of miR-375 (Additional file 7: Figure S4C).

Finally, we determined whether forced overexpression of miR-375 inhibits the leukemogenesis in a nude mouse xenograft model. HL-60 cells were transduced with MSCV-NC or MSCV-miR-375, followed by puromycin selection for 1 week. Nude mice were divided in two groups (N = 6 per group) and inoculated with HL-60-miR-375 or HL-60-NC. Tumors in mice inoculated with HL-60-miR-375 were considerably smaller than those in mice inoculated with negative control (Fig. 7a). Furthermore, tumor growth was suppressed in mice inoculated with HL-60-miR-375 than control mice (Additional file 8: Figure S5A). Similarly, the average tumor volume was reduced by 34.7% (Fig. 7b) and average tumor weight was reduced by 38.5% (Fig. 7c) in HL-60-miR-375-inoculated mice compared with control mice. Consistent with the results from cell lines, the protein levels of HOXB3 were considerably decreased in mice inoculated with HL-60-miR-375 compared with control mice (Fig. 7d).

To further examine the effects of miR-375 overexpression in vivo, THP1 cells were transduced with a lentivirus encoding enhanced GFP followed by flow cytometric sorting for GFP-positive cells, which were further transduced with MSCV-NC or MSCV-miR-375. THP1-miR-375 or THP1-NC cells were xenografted into nonobese diabetic scid gamma (NSG) mice. We detected GFP cells in peripheral blood when mice transplanted with THP1-NC became moribund. The percentage of GFP-positive cells was reduced by 35% in mice transplanted with THP1-miR-375 than in mice transplanted with THP1-NC (Fig. 7e). We then evaluated whether miR-375 reduced the infiltration of blast cells in the spleen. The spleen volume and weight were reduced in THP1-miR-375-transplanted mice than in THP1-NC-transplanted mice (Fig. 7f). Extensive infiltrations of immature blast cells destroyed the normal architectures in spleens from THP1-NC-transplanted mice (Additional file 8: Figure S5B, Left). Accordingly, forced expression of miR-375 substantially decreased the infiltrations (Additional file 8: Figure S5B, Right). Finally, miR-375 significantly extended the survival time in mice (P < 0.01; Fig. 7g).