Reintroduction of CEBPA in MN1-overexpressing hematopoietic cells prevents their hyperproliferation and restores myeloid differentiation

The decision between self-renewal and lineage commitment of hematopoietic stem cells is mainly regulated by extracellular signals from the hematopoietic microenvironment that induce differential expression of key transcription factors. Aberrant expression or function of these factors contributes significantly to leukemogenesis.^1,2  Clonal expansion of immature hematopoietic cells resulting from alterations of differentiation-inducing transcription factors is the hallmark of acute myeloid leukemia (AML).^2,3 

MN1, meningioma 1, was identified in a patient with meningioma, as a target of a chromosome translocation t(4;22).^4,5  Subsequent studies showed that MN1 is also involved in AML either as a partner of a recurrent translocation t(12;22)(p12;q12), encoding an MN1-TEL fusion protein,⁶  or as an overexpressed gene in 2 subsets of AML specified by the presence of inv(16) or by overexpression of ecotropic viral integration site 1.^7,8  Our recent studies showed that ectopic expression of MN1 or MN1-TEL in hematopoietic cells causes myeloid disease in mice.^9,10  Moreover, MN1 cooperates with CBFβ-MYH11, the product of inv(16), in a murine model of inv(16)-AML and accelerates leukemogenesis.⁹  MN1 also prevents granulocytic differentiation and abrogates all-trans retinoic acid (ATRA)–mediated growth inhibition in primary mouse hematopoietic cells.¹¹  In addition, MN1 overexpression eliminated the beneficial effects of the addition of ATRA to the maintenance chemotherapy of nonacute promyelocytic leukemia–AML patients; whereas in AML patients with low MN1 expression, ATRA administration increased survival.¹¹ 

The transcription factor CEBPA, CCAAT enhancer binding protein-α (C/EBPα), plays an essential, nonredundant role in the decision of hematopoietic stem cells to self-renew or differentiate into a myeloid direction¹²  and it is activated by ATRA.¹³  CEBPA regulates expression of several myeloid-specific genes, such as granulocyte-colony stimulating factor receptor (G-CSFR), monocyte-colony stimulating factor receptor (M-CSFR),¹⁴  and miR-223¹³  via direct binding to sites in their promoters. Cebpa knockout mice showed a profound defect in the generation of granulocyte-monocyte progenitors from common myeloid progenitors, resulting in a lack of granulocytes and impaired macrophage development.^12,14,15  Concordantly, the absence of Cebpa enhances the repopulating and self-renewal capacity of hematopoietic stem/progenitor cells (HSPCs),^12,15  whereas its forced expression in human CD34⁺ primary cells and AML cells induced myeloid differentiation.^16,17  It is therefore logical that the loss of CEBPA function is frequently observed in AML.^18-21 

Although MN1 is a novel player in AML pathophysiology, the molecular mechanisms via which MN1 inhibits differentiation and stimulates proliferation of hematopoietic cells remain elusive. Here, we expanded on previous studies and addressed the following questions: (1) Is monocytic differentiation protected in MN1 cells at the expense of granulocytic differentiation? (2) Are the effects of MN1 on primary hematopoietic cells similar to those in human AML cell lines? (3) How does the transcriptome of primary human CD34⁺-MN1 cells differ from that of parental CD34⁺ cells? To answer these questions, we used cell culture experiments to analyze the effects of MN1 overexpression on the proliferation and myeloid differentiation of human and mouse primary HSPC and AML cell lines. To gain insight into molecular mechanisms orchestrating the MN1-induced phenotype, we performed microarray analysis using RNA from control and MN1-overexpressing CD34⁺ hematopoietic cells. Furthermore, we determined the effects of reintroduction of conditional and constitutive CEBPA on the MN1-mediated inhibition of differentiation and stimulation of proliferation because CEBPA expression was repressed in CD34⁺-MN1 cells.

Human CEBPA (Origene) and MN1 cDNAs²²  were cloned into MSCV-IRES-GFP (MIG) or MSCV-IRES-YFP (MIY) retroviral vectors as Xho/EcoRI and EcoRI fragments, respectively. A tamoxifen inducible MIGR1-CEBPA-ER retroviral vector was kindly provided by Dr Alan D. Friedman (Johns Hopkins University, Baltimore, MD). We generated VSVg-pseudotyped retrovirus by transient transfection of 293T cells maintained in Dulbecco modified Eagle medium supplemented with 10% fetal bovine serum (FBS; Hyclone; Thermo Fisher Scientific) Glutamax (2 mM) and penicillin/streptomycin (100 U/mL each, Invitrogen; referred to as complete medium). For virus production, we used complete medium supplemented with 10% heat inactivated FBS.

After approval by the Institutional Review Board of St Jude Children's Research Hospital, bone marrow (BM) or RNA samples were obtained from AML patients and mononuclear cells were separated via Ficoll-gradient centrifugation of apheresis samples of G-CSF–mobilized transplantation donors. Informed consent was provided according to the Declaration of Helsinki. Monocytes, granulocytes, and CD34⁺ cells were isolated from mononuclear cells, expanded, and differentiated using existing protocols.²³  Purity of isolated cells was confirmed by fluorescence-activated cell sorting (FACS) analysis. CD34⁺ cells (2-4 × 10⁵ cells/1 mL per well of retronectin-coated 24-well plates) were transduced 3 times during a 2-day interval with MIG or MIG-MN1 virus prepared in growth medium (serum-free X-VIVO15, BioWhittaker, 04-744Q; supplemented with Glutamax and penicillin/streptomycin), diluted 1:1, and supplemented with progenitor cytokines (human stem cell factor [SCF], fms-related tyrosine kinase 3 ligand, and thrombopoietin; 50 ng/mL each).²³  For in vitro differentiation assays, 2- to 3-day expanded CD34⁺ cells were plated (2-3 × 10⁴ cells/mL) in growth medium containing monocyte cytokines: macrophage colony-stimulating factor (M-CSF; 100 ng/mL), SCF (20 ng/mL), interleukin-6 (IL-6; 20 ng/mL), fms-related tyrosine kinase 3 ligand (50 ng/mL), ± vitamin D₃ (60 nM; Sigma-Aldrich) to induce monocytic differentiation; or granulocyte cytokines [G-CSF (100 ng/mL) and SCF (20 ng/mL; all cytokines from PeproTech) to induce granulocytic differentiation.²³  Half of the medium was replaced every 3 days, and differentiation was assessed on days 10 to 14 of culture using FACS analysis.

Human BM-derived CD34⁺ cells (StemCell Technologies) were expanded and transduced with mock or MN1 retrovirus. One day later, RNA was isolated from GFP⁺/CD34⁺ FACS-sorted cells using Trizol (Sigma-Aldrich) and was subjected to microarray analysis following Affymetrix protocols (Affymetrix) using the GeneChip Human U133 Plus 2.0 array. The microarray data for this study have been deposited into GEO under accession no. GSE16745.²⁴ 

Methyl-thiazol-tetrazolium (MTT) assay was performed following the manufacturer's instructions (Promega). Methylcellulose assays were performed by plating FACS-sorted, engineered cells into MethoCult GF H4444 (StemCell Technologies). Primary AML cells were expanded for 1 day and transduced using the progenitor cytokines and the same conditions as those used for primary CD34⁺ cultures.

HL60 and U937 cells (2.5 × 10⁵/well, 12-well plate) were transduced with 2 mL of the corresponding MIG or MIY retrovirus 1:1 diluted in complete RPMI containing 10% heat-inactivated FBS in the presence of 8 μg/mL polybrene (Chemicon) for 2 days (total 2 or 3 times). GFP⁺ or YFP⁺ cells were sorted using FACS and induced to differentiate with different doses of vitamin D₃ (Sigma-Aldrich, D1530), ATRA (Sigma-Aldrich, R2625), or the same volume of vehicle for 2 to 3 days. Differentiation was assessed using FACS analysis and May-Grünwald-Giemsa staining of cytospin preparations. CEBPA-ER-expressing cells were additionally treated with 4-hydroxytamoxifen (Sigma-Aldrich, H7904) or vehicle for the indicated times and doses.

Total RNA was extracted using Trizol or the mirVana miRNA isolation kit (Ambion) following the manufacturer's protocol. cDNA was synthesized using hexamer primers and Superscript II reverse transcriptase (Invitrogen). Primer and probes for human HPRT, GAPDH, RARβ, CEBPA, p21, G-CSFR, M-CSFR, and CALR were from Applied Biosystems (assay ID: 4326321E, 4310884E, Hs00977137-m1, Hs00269972-s1, Hs99999142-m1, Hs00167918-m1, Hs00911250-m1, Hs00189032-m1, respectively). The MN1 primer and probe sets were as described.⁹  Expression levels were obtained using the standard curve method (after normalization with a housekeeping gene). The TaqMan microRNA assay kit for RNU6B (used as endogenous control and normalizer) and miR-223 were from Applied Biosystems (part no. 4373381 and 4373075, respectively). Reverse-transcribed polymerase chain reaction (RT-PCR) reactions were performed according to the manufacturer's instructions.

Cells were washed and taken up in blocking solution (phosphate-buffered saline containing 5% fetal bovine serum, 2 mM sodium azide, and 0.1 mg/mL human gamma globulin) for 30 minutes on ice. After washing, cells were stained with the indicated directly conjugated antibodies for 30 minutes on ice, washed, and analyzed using an LSR II flow cytometer (BD Biosciences). Anti–human CD11b-allophycocyanin (APC), CD14-phycoerythrin (PE), CD15-APC, CD45-APC, CD34-PE, and CD38-APC were from Miltenyi Biotec, and anti–mouse antibodies (Mac1-PE, Gr1-APC, Sca1-PerCP-Cy5-5) were from BD Biosciences, except for ckit-APC-Alexa750 (eBioscience). In some experiments, cells engineered to express either GFP or YFP or both were counterstained with propidium iodide to mark dead cells and were analyzed using FACS Vantage-SE-DiVa cell sorters (BD Biosciences); subpopulations of viable cells expressing these fluorescent proteins were collected by FACS. Cell cycle and apoptosis assays were performed using FACS as described.²⁵ 

Total cell lysates were prepared by sonication for 15 seconds after incubating the cells on ice in radio immunoprecipitation assay buffer (50 mM Tris, pH 7.5, 150 mM NaCl, 1% deoxycholate, 1% Triton X-100, 0.1% sodium dodecyl sulfate) containing protease and phosphate inhibitors (Sigma-Aldrich). Lysates were subjected to Western blot analysis to detect CEBPA (39306; active motif) and GAPDH (MAB374; Millipore) expression following the manufacturer's instructions. Immunocytochemistry of HL60-MN1 and U937-MN1 cells using anti-MN1 antibody was performed as described.⁹  Images of cytospins were captured with an Olympus BX-50 microscope (equipped with UPlanFL 40×/0.75 and UPlanFL 60×/1.25 numerical aperatures with a SPOT camera and SPOT Advanced Imaging software (Diagnostic Instruments). The original magnification for cytospins was ×400 or ×600.

BM cells were isolated and cultured as described by Wang et al,²⁶  with additional lineage depletion after initial BM isolation. Methylcellulose assays were performed by plating FACS-sorted, engineered cells into MethoCult GF M3534 (StemCell Technologies).²⁷ 

Unpaired t test analysis (2-tailed) was performed using the GraphPad Prism, version 4.0c for Mac (GraphPad Software; www.graphpad.com).

Forced expression of MN1 in mouse hematopoietic cells prevented granulocytic differentiation, abrogated ATRA-mediated growth inhibition, and caused myeloid disease in mouse models.^9,11  To further determine the effects of MN1 on myeloid differentiation of human AML cell lines, HL60 and U937 cells were transduced with MIG or MIG-MN1 retroviruses, respectively. Immunocytochemistry and real-time RT-PCR analysis of GFP⁺ FACS-sorted cells showed MN1 expression in the nucleus (Figure 1A bottom panel) and similar levels of MN1 expression in the MN1-transduced cell lines, whereas mock-transduced HL60 and U937 cells did not express endogenous MN1 (Figure 1B).

Compared with mock-transduced cells, exogenous MN1 inhibited ATRA-induced growth arrest of HL60 cells by 2-fold (Figure 1C right panel) and differentiation, as monitored by the induction of granulocytic markers (CD11b⁺/CD15⁺ or CD11b) in both HL60-MN1 (from 62% to 16%) and U937-MN1 cells (from 71% to 27%; Figure 1D,F). Although vitamin D₃-mediated growth arrest was not inhibited significantly (Figure 1C middle panel), vitamin D₃-induced monocytic differentiation was also inhibited in both MN1-transduced cell lines (Figure 1E-F). FACS analysis indicated that vitamin D₃-induced expression of CD11b was decreased by 2.6-fold (P = .005) and 1.8-fold (P = .001), and CD14 expression by 4-fold (P = .004) and 3.2-fold (P < .001) in HL60-MN1 cells treated with 10 nM and 100 nM vitamin D₃, respectively (mock and MN1, Figure 1E). Similar vitamin D₃ effects were seen in U937-MN1 cells with a 2-fold reduction in CD11b expression (vitamin D₃, U937-mock, and U937-MN1, Figure 1F). Thus, MN1 overexpression impaired not only granulocytic but also monocytic differentiation of AML cell lines in response to ATRA and vitamin D₃. To determine the relevance to AML patients who overexpress MN1, we sorted U937-MN1 cells into a low, a medium, and a high GFP-expressing fraction and compared the amount of MN1 RNA in these fractions with that in CD34⁺ BM cells and in 6 different inv(16) AML patient samples that overexpress MN1. Expression in the low fraction of U937-MN1 cells was 3.4-fold higher than in the highest expressing patient sample (Figure 1F top right panel), whereas the medium and high fractions are 5-fold and 7.4-fold higher, respectively. Curiously, the inhibition of CD11b expression by MN1 is the strongest in the low fraction (vitamin D₃, 2.1-fold; ATRA, 3.6-fold, Figure 1F) slightly less effective in the medium (vitamin D₃, 2.4-fold; ATRA, 3.3-fold) and least effective in the high fraction (vitamin D₃, 1.3-fold; ATRA, 1.4-fold; blue, magenta, and orange in Figure 1F bottom panel, U937-mock and U937-MN1). This suggests that disproportionate overexpression of MN1 actually impairs its differentiation-inhibiting effect, which bodes well for the effects mediated by the lower MN1 overexpression levels in patient samples.

We next analyzed endogenous MN1 expression in RNA samples obtained from human BM-derived CD34⁺ cells as well as from primary peripheral blood monocytes and granulocytes of G-CSF-mobilized healthy donors. Compared with CD34⁺ cells, the level of endogenous MN1 mRNA was down-regulated in both monocytes (17- to 86-fold) and granulocytes (216- to 370-fold; Figure 2A). In an independent experiment, CD34⁺ cells were maintained in cultures supporting HSPC growth for 5 days followed by differentiation for 14 days in the presence of monocyte or granulocyte cytokine cocktails.²³  FACS analysis confirmed that CD34⁺ cells maintained their identity in the presence of progenitor cytokines (Figure 2B), whereas the presence of G-CSF or M-CSF induced differentiation and expression of CD15 (92%) and CD14 (57%), respectively (Figure 2B). Similar to uncultured peripheral blood monocytes and granulocytes, endogenous MN1 expression was down-regulated in both in vitro-differentiated monocytes (8.3-fold; CD14 graph, Figure 2B) and granulocytes compared with cultured CD34⁺ cells (92-fold; CD15 graph, Figure 2B). Collectively, these results suggested that the endogenous MN1 expression must have been switched off during the differentiation process.

We next analyzed whether MN1 expression also promoted proliferation and inhibited differentiation of primary human CD34⁺ cells. We transduced CD34⁺ cells with mock or MN1-expressing MIG retroviruses and 2 days later induced differentiation of unsorted cells in liquid cultures containing a monocyte-cytokine cocktail in the presence or absence of vitamin D₃.²³  Coculturing of normal and transduced cells in this experiment provided an internal control for the culture conditions, as well as a test for possible autocrine/paracrine effects between untransduced and transduced cells. FACS analysis showed that, in the mock-transduced sample, M-CSF treatment resulted in a similar percentage of CD14-expressing cells in both the GFP⁻ (39%) and GFP⁺ (37%) cells, and this percentage was increased 2.2-fold in the presence of vitamin D₃ (Figure 2C). Although M-CSF and M-CSF + vitamin D₃ treatment of the MN1-transduced culture resulted in a similar percentage of CD14-expressing GFP⁻ cells, the MN1-expressing (GFP⁺) cells displayed a 3.6-fold and 2.5-fold drop in CD14 expressing cells, respectively (Figure 2C). Indeed, cytospin preparations of GFP⁺ FACS-sorted mock cells from M-CSF cultures displayed different levels of monocytic differentiation, whereas MN1 cells showed mostly blast morphology (Figure 2C top panel). Reduced sensitivity to M-CSF in CD34⁺-MN1 cells correlated with decreased expression of M-CSFR (Figure 3A). Similar results were obtained in an independent experiment in which we cocultured FACS-sorted GFP⁺ and GFP⁻ cells at a 1:1 ratio in a monocytic differentiation assay (data not shown). Therefore, forced MN1 expression impaired the monocytic differentiation of human CD34⁺ cells in vitro through cell-autonomous mechanisms.

To determine whether the inhibitory effect of MN1 on differentiation was accompanied by increased proliferation, sorted (GFP⁺) or unsorted mock and MN1-transduced human CD34⁺ cells were maintained in progenitor media, and the proliferation capacity of the cells was assessed using the MTT assay (at day 0, 2, 4, and 6; Figure 2D) or by performing periodic GFP analysis at 3, 12, and 30 days after transduction, respectively. Sorted (CD34⁺/GFP⁺) cells were also subjected to serial replating in methylcellulose assays to determine the effect of MN1 on self-renewal. As in MN1-transduced mouse HSPCs,⁹  ectopic MN1 expression increased the self-renewal activity (Figure 2F) and proliferation rate of CD34⁺ cells as the percentage of GFP⁺ cells increased from 7% at day 3 to 76% at day 30, whereas it remained stable (16%-28%) in the mock culture (Figure 2E). In a separate experiment, FACS-sorted mock and MN1 cells were cultured in the presence of progenitor cytokines for 10 days and subjected to FACS analysis to determine the cell cycle status and survival of the cells. Annexin-V analysis showed 55% viability in mock cells, whereas it was increased to 89% in MN1 cells (Figure 2G). MN1 cells also showed decreased G₀/G₁ (MN1, 60%; mock, 79%, P = .002) and an increased S/G₂M ratio (MN1, 40%; mock, 21%, P < .01; Figure 2H). Together, these data suggested that ectopic MN1 enhanced the self-renewal and proliferation of human CD34⁺ cells via increased cell cycle traverse and survival in vitro, which was accompanied by impaired differentiation.

To gain insight into molecular mechanisms orchestrating the MN1-induced effects, we performed microarray analysis of RNA samples of human BM-derived FACS-sorted CD34⁺ cells transduced with mock or MN1-expressing retroviruses (CD34⁺/GFP⁺). Among the many differentially expressed genes, MN1 cells showed decreased expression of CEBPA (2.1-fold) and of 2 of its downstream targets: miR-223¹³  (3.5-fold) and p21²⁸  (1.9-fold; data not shown). Real-time RT-PCR analysis using RNA from G-CSF mobilized peripheral blood CD34⁺ cells showed similarly decreased levels of CEBPA (10-fold), and its downstream targets p21 (5.3-fold), G-CSFR (4.8-fold), and M-CSFR (2.6-fold) in MN1 cells, whereas expression of calreticulin (CALR), which regulates CEBPA expression at the translational level,²⁹  was moderately (1.3-fold) but significantly (P = .009) affected (Figure 3A).

Our microarray data suggested that the down-regulation of CEBPA might contribute to the observed lack of differentiation and increased proliferative effects of MN1 in hematopoietic cells. To test this hypothesis, we first analyzed the expression of CEBPA and its direct targets in response to ATRA, which normally induces the expression of CEBPA targets.¹³  U937-mock or MN1 cells were treated with 1 μM ATRA for 4, 8, 16, and 24 hours, and expression of CEBPA or G-CSFR was determined using real-time RT-PCR. CEBPA was slightly but significantly down-regulated at 8 and 24 hours in untreated or vehicle-treated MN1 cells compared with mock cells (Figure 3B). Addition of ATRA slightly but significantly (P = .003) increased CEBPA expression at 8 hours in mock cells compared with vehicle-treated mock cells. Although a similar increase was observed after 8 hours of ATRA treatment of MN1 cells compared with vehicle-treated cells, the increase was still significantly lower (P = .042) than in ATRA-treated mock cells at the same time points. ATRA treatment (24 hours) also induced a gradual 3.5-fold increase in G-CSFR expression in U937-mock cells (Figure 3C), which was accompanied by increased expression of the differentiation marker CD11b (Figure 3D). In contrast, G-CSFR expression in MN1 cells was maintained at the level of that in untreated mock cells during ATRA treatment (Figure 3C), and CD11b expression was 2.8-fold lower in MN1 cells treated for 24 hours compared with treated mock cells (Figure 3D). Up-regulation of miR-223, another direct target of CEBPA that is critical for granulocytic differentiation,¹³  was also partially blocked in ATRA-treated U937-MN1 cells (Figure 3E).

We also analyzed the expression of RARβ, an ATRA-activated RAR/RXR target, in the same RNA samples in Figure 3C. As expected, untreated U937-mock cells did not express RARβ, whereas mock and MN1 cells showed similar levels of RARβ induction after 4, 8, and 24 hours of ATRA treatment (Figure 3F). In a separate experiment, the RARβ level was also determined in 24-, 48-, and 72-hour ATRA-treated HL60-mock or MN1 cells and similar results were obtained (Figure 3G). We concluded that MN1 overexpression moderately down-regulates CEBPA expression in U937 cells; and in response to ATRA, activation of the CEBPA target genes, G-CSFR and miR-223, was impaired in MN1 cells, whereas RAR/RXR activity was preserved.

Given the effects of forced MN1 expression on CEBPA expression and its target genes in CD34⁺ cells, we reasoned that reintroduction of CEBPA might bypass the effects of MN1 on differentiation and proliferation. Therefore, YFP⁺-sorted U937-mock and MN1 cells were transduced with MIG retrovirus expressing human CEBPA, and the proliferation of YFP⁺/GFP⁺-sorted cells was analyzed. CEBPA overexpressing mock and MN1 cells showed a 2.7-fold decrease in growth rate at day 4 of this analysis (Figure 4A), which was accompanied by differentiation as May-Grünwald-Giemsa–stained cytospin preparations displayed morphologic features of differentiated cells, such as segmented nuclei, in mock + CEBPA and MN1 + CEBPA cells but not in the respective parental cell lines (Figure 4B). Moreover, FACS analysis showed that compared with the parental cells, CD11b was induced in 55% of CEBPA-overexpressing mock and MN1 cells (Figure 4C). Additional real-time RT-PCR analysis of RNA samples obtained from day 4 cells showed a 2-fold increase in G-CSFR expression in both mock + CEBPA and MN1 + CEBPA cells compared with controls (Figure 4D).

To determine whether ATRA or vitamin D₃ treatment could further enhance the differentiation of MN1 + CEBPA cells, YFP-expressing U937-mock or MN1 cells were transduced with tamoxifen-inducible MIGR1-CEBPA-ER retrovirus.²⁶  Treatment of mock and MN1 cells with 10 or 50 nM tamoxifen for 2 days showed a dose-dependent induction of differentiation markers (Figure 4F). In a separate experiment, cells were treated with vehicle or tamoxifen (100 nM) for 2 days in the presence or absence of ATRA and vitamin D₃, respectively. FACS analysis of GFP⁺/YFP⁺ cells showed that expression of CD11b was induced to a similar degree (81% vs 84%) in tamoxifen and tamoxifen + ATRA-treated CEBPA-ER cells (mock + CEBPA-ER), and cotreatment with tamoxifen + ATRA further enhanced the expression level of CD11b in MN1 + CEBPA-ER cells compared with cells treated with tamoxifen alone (77% and 62%, respectively; Figure 4G). Tamoxifen + vitamin D₃ treatment synergistically increased the expression of the monocytic marker CD14 compared with tamoxifen alone treated cells in MN1 + CEBPA-ER (20-fold increase) as well as in mock + CEBPA + ER cells (8-fold increase; Figure 4H).

To study the differentiation and proliferation capacity of primary mouse HSPC-expressing MN1 or MN1 + CEBPA-ER, lin⁻ BM cells were transduced sequentially with YFP-expressing mock or MN1 retroviruses and GFP-expressing CEBPA-ER retroviruses (Figure 5A diagram). FACS profiles of transduced cells after a second lineage depletion (this time point referred as day 0), which eliminated the cells that committed during the transduction procedure,²⁶  showed that the majority of the untreated day 0 cells were c-Kit⁺ or Sca1⁺/c-Kit⁺ positive progenitor/stem cells (Figure 5B). These cells were cultured with vehicle or tamoxifen for 2 days in the presence of SCF, IL-3, and IL-6, and myeloid differentiation during this culture period was monitored using FACS. At day 2, the percentage of Mac1⁺/Gr⁺ or Gr1⁺ positive granulocytes had increased from 22% (Figure 5C) to 69% (Figure 5D) in vehicle-treated YFP⁺/GFP⁻-mock cells. In contrast, granulocytic differentiation was inhibited in untreated day 0 (Figure 5C) and vehicle treated day 2 MN1 cells (Figure 5D) by 2.6- and 4.3-fold, respectively. The percentage of Mac1⁺/Gr1⁻ monocytic cells was also reduced by ectopic MN1 expression at untreated day 0 and vehicle-treated day 2 cells, by 3-fold and 1.7-fold, respectively (Figure 5C-D). The MN1 effect on myeloid differentiation was eliminated when CEBPA-ER was reintroduced and activated in MN1 cells, as judged by increased numbers of Mac1⁺/Gr1⁺ granulocytes and Mac1⁺ monocytes (Figure 5E).

In a separate experiment, we analyzed whether the increased proliferation of ectopic MN1-expressing primary hematopoietic cells could be prevented by reintroduction of CEBPA. We cotransduced lin⁻ mouse BM cells with GFP and YFP-retroviruses, sorted GFP⁺, YFP⁺, and GFP⁺/YFP⁺ cells by FACS (mock, MN1, constitutive human CEBPA, MN1 + CEBPA), and plated the cells into methylcellulose media supporting myeloid colony formation. Similar to our previous data, MN1 cells showed up to 3-fold increase in colony formation, reflecting the increased numbers of HSPCs.⁹  The increased colony formation of MN1 cells was lost when CEBPA was reintroduced into MN1 cells (Figure 5F), which is consistent with the observed growth arrest/differentiation of U937-MN1 + CEBPA cells (Figure 4A). Therefore, in primary HSPC and U937 cells, forced CEBPA expression was sufficient to override the MN1-induced partial block in myeloid differentiation and to suppress MN1-mediated proliferation.

To determine whether ectopic CEBPA expression could also induce differentiation of MN1-overexpressing primary AML cells, we transduced 2 AML patients' BM samples (patients 5 and 6, Figure 1F) exhibiting 94% and 89% blast, respectively, with MIG or MIG-CEBPA retroviruses. FACS analysis showed that CEBPA overexpression increased the level of CD14 by 2-fold in patient 5 and 1.3-fold in patient 6, whereas the expression of CD11b showed only very moderate change (Figure 6). This indicates that CEBPA forces myeloid differentiation in AML patients who overexpress MN1.

Our main conclusions of this study are as follows: (1) Ectopic MN1 expression not only inhibited granulocytic differentiation¹¹  but also vitamin D₃, M-CSF, or growth factor-induced monocytic differentiation of human CD34⁺ cells, primary mouse lin⁻ hematopoietic cells, and AML cell lines in vitro. (2) Compared with CD34⁺ cells, endogenous MN1 expression is down-regulated on monocytic/granulocytic differentiation and forced expression of MN1 in human CD34⁺ cells down-regulated CEBPA expression, increased their self-renewal, proliferation, survival, and cell cycle traverse in response to hematopoietic growth factors in vitro. (3) Reintroduction of tamoxifen-inducible or -constitutive CEBPA bypassed these effects of MN1, suggesting that CEBPA repression contributes to the MN1-induced hematopoietic phenotype.

Forced MN1 expression in lin⁻ mouse BM cells enhanced their self-renewal capacity, resulting in the generation of cell lines and the induction of myeloid disease in a mouse model.^9,11  Here we showed that enforced expression of MN1 in human CD34⁺ hematopoietic cells also caused enhanced self-renewal (Figure 2F) and proliferative advantage of MN1 cells over that of control mock cells in liquid cultures promoting the growth of HSPCs (Figure 2D-E). Furthermore, subsequent analysis of CD34-mock and CD34-MN1 cells demonstrated that this was the result of increased survival and cell cycle traverse of MN1 cells (Figure 2G-H, respectively). We think that this activity is cell type and context dependent, given that, in human osteoblast³⁰  and epithelial cell lines,³¹  increased MN1 expression is correlated with growth arrest.

Although we do not know how MN1 induces growth of hematopoietic cells, we think that direct or indirect repression of CEBPA and its target genes (microarray analysis, Figure 3A; and data not shown), all critical regulators of self-renewal/proliferation and lineage commitment,^12,16  is one of the important consequences of MN1 overexpression. Our findings raise the possibility that MN1 might be involved in self-renewal/myeloid commitment decisions of primitive hematopoietic cells, which is further supported by the following observations: (1) Endogenous MN1 expression is down-regulated in primary or in vitro differentiated human monocytes and granulocytes (Figure 2A-B). (2) Microarray analysis of cord blood-derived human CD34⁺ hematopoietic cells showed higher MN1 expression in the CD34⁺/CD38⁻ fraction (highly enriched in long-term engrafting cells³² ), than in the committed CD34⁺/CD38⁺ cells,³³  and knockdown of CEBPA inhibits myeloid differentiation of human CD34⁺ cells.³⁴  (3) MN1 and CEBPA exhibit inversed expression patterns in primitive cells; CEBPA expression is lower in hematopoietic stem cells and higher in more committed cells.³⁵ 

AML is characterized by unlimited self-renewal activity and impaired differentiation of hematopoietic cells, which results in accumulated myeloid blasts in the BM and periphery.³  Enforced expression of MN1 fulfills these criteria by deregulating both activities^9,11  (Figures 1C-D, 2C,E,F, and 5E-F). Given that CEBPA is down-regulated in CD34-MN1 cells (Figure 3A) and is indispensable for granulocytic and monocytic maturation,^15,26  we first analyzed whether CEBPA expression is also down-regulated in U937-MN1 cells. However, ectopic MN1 only moderately down-regulated endogenous CEBPA expression at the transcriptional level in U937 cells (Figure 3B), and whether MN1 directly represses CEBPA remains elusive. Although CEBPA expression only changed moderately, ATRA-induced activation of the CEBPA target genes, G-CSFR and miR223, was impaired in U937-MN1 cells, whereas activation of RAR/RXR, another well-established transcription factor regulating myeloid differentiation, was not altered as judged by similar expression of its direct target RARβ (Figure 3F-G). This is consistent with our previous findings showing that MN1 and ATRA synergistically activated RAR/RXR-induced transcriptional activity in Hep3B cells.³⁶  This makes a stronger case for functional involvement of MN1-mediated repression of CEBPA in the inhibition of differentiation of hematopoietic cells. Supporting this notion, we found that reintroduction of constitutive human CEBPA or tamoxifen-inducible rat Cebpa²⁶  bypassed the inhibitory effect of MN1 and promoted both monocytic and granulocytic differentiation in U937 cells, mouse lin⁻ BM cells, and primary AML cells that overexpress MN1 (Figures 4,Figure 5–6).

Currently, we do not know how MN1 exerts its effect on CEBPA expression and activity, but our data clearly indicate that reintroduction of CEBPA is sufficient to limit the effect of MN1 on proliferation and myeloid differentiation in vitro. This is similar to CEBPA's effect on human wild-type CD34⁺ and AML-CD34⁺ cells.¹⁷  Based on our results, it is tempting to speculate that, in hematopoietic cells, MN1 represses CEBPA expression via factors required for its transcription, or perhaps via epigenetic modifications of the CEBPA promoter region. CEBPA exerts its function via direct binding to promoter/enhancer regions of its target genes as well as via posttranslational modifications and protein-protein interactions.^29,37  Because MN1 can bind the transcriptional coactivator p300,³⁶  we analyzed whether recruitment of MN1 to the CEBPA/p300 protein complex would inhibit CEBPA's transcriptional activity in U937-MN1 cells, but we were unable to detect any interaction between CEBPA and MN1 (data not shown). Furthermore, the amount of p300 bound to CEBPA remained unaltered (data not shown). We cannot exclude that binding of other proteins/coactivators to CEBPA might be impaired in MN1 cells. Therefore, future studies, including detailed identification of MN1 interacting proteins, might provide clues about the molecular consequences of MN1 expression.

The authors thank Dr Alan D. Friedman for kindly providing MIGR1-CEBPA-ER construct; Dr Mihaela Onciu for the analysis of cytospin preparations; Yvan Campos for subcloning of CEBPA; Dr Youngsoo Lee for providing help for real-time RT-PCR analysis; Dr Richard Ashmun, Dr Geoff Neale, Martha Holladay, and members of FACS and microarray facility for expert data analysis; Drs Jan Jacob Schuringa, Thasia Leimig, Hakan Cam, and John Easton for helpful suggestions; Sheila Shurtleff for providing patient cells; the St Jude blood donor center for providing apheresis rings; and Charlette Hill for secretarial assistance.

This work was supported by the National Cancer Institute (grant CA72999), the Cancer Center (support grant CA021765), and the American Lebanese Syrian Associated Charities.

Contribution: A.K. designed and performed research, analyzed data, and wrote the paper; and G.C.G. designed research and wrote the paper.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Gerard C. Grosveld, Department of Genetics and Tumor Cell Biology, St Jude Children's Research Hospital, 262 Danny Thomas Place, Memphis, TN 38105; e-mail: gerard.grosveld@stjude.org.