Runx1 Is Expressed in Adult Mouse Hematopoietic Stem Cells and Differentiating Myeloid and Lymphoid Cells, But Not in Maturing Erythroid Cells

Authors

  • Trista E. North,

    1. Department of Biochemistry, Dartmouth Medical School, Hanover, New Hampshire, USA
    Search for more papers by this author
  • Terryl Stacy,

    1. Department of Biochemistry, Dartmouth Medical School, Hanover, New Hampshire, USA
    Search for more papers by this author
  • Christina J. Matheny,

    1. Department of Biochemistry, Dartmouth Medical School, Hanover, New Hampshire, USA
    Search for more papers by this author
  • Nancy A. Speck,

    1. Department of Biochemistry, Dartmouth Medical School, Hanover, New Hampshire, USA
    Search for more papers by this author
  • Dr. Marella F.T.R. de Bruijn Ph.D.

    Corresponding author
    1. Department of Biochemistry, Dartmouth Medical School, Hanover, New Hampshire, USA
    Current affiliation:
    1. MRC Molecular Haematology Unit, Weatherall Institute of Molecular Medicine, John Radcliffe Hospital, Oxford, United Kingdom
    • MRC Molecular Haematology Unit, Weatherall Institute of Molecular Medicine, John Radcliffe Hospital, Oxford OX3 9DS, United Kingdom. Telephone: 44-1865-222397; Fax: 44-1865-222500
    Search for more papers by this author

Abstract

The transcription factor Runx1 marks all functional hematopoietic stem cells (HSCs) in the embryo and is required for their generation. Mutations in Runx1 are found in approximately 25% of acute leukemias and in familial platelet disorder, suggesting a role for Runx1 in adult hematopoiesis as well. A comprehensive analysis of Runx1 expression in adult hematopoiesis is lacking. Here we show that Runx1 is expressed in functional HSCs in the adult mouse, as well as in cells with spleen colony-forming unit (CFU) and culture CFU capacities. Additionally, we document Runx1 expression in all hematopoietic lineages at the single cell level. Runx1 is expressed in the majority of myeloid cells and in a smaller proportion of lymphoid cells. Runx1 expression substantially decreases during erythroid differentiation. We also document effects of reduced Runx1 levels on adult hematopoiesis.

Introduction

Runx1 belongs to the small family of core-binding factors (CBFs). Runx1 binds DNA through a conserved sequence, the Runt domain, and forms together with the non-DNA binding CBFβ subunit, the heterodimeric transcription factor Runx1-CBFβ. Runx1-CBFβ has been found to play a crucial role in definitive hematopoiesis during ontogeny [1]. The high frequencies of RUNX1 and CBFB mutations in acute leukemias and myelodysplasia and the demonstration that haploinsufficiency of RUNX1 causes familial platelet disorder suggest a role for this factor in postnatal hematopoietic differentiation as well [1, 2]. Potential Runx1-CBFβ target genes have been identified in the erythroid, myeloid, and lymphoid lineages [3]. Runx1 can either activate or repress transcription, depending on which other proteins bind to Runx1 or adjacent sites in target promoter/enhancer regions [4]. Although both Runx1 and CBFβ are expressed in adult bone marrow (BM), thymus, and peripheral lymphoid organs [513], a detailed characterization and functional analysis of Runx1-expressing cells in adult hematopoiesis has been lacking. Comprehensive mapping of Runx1 expression in hematopoietic stem and progenitor cell populations would facilitate functional studies of the role of Runx1 in normal and aberrant adult hematopoiesis.

We previously generated a Runx1 allele in which two coding exons were replaced with lacZ coding sequences (Runx1lz) [14]. The Runx1-β-galactosidase fusion protein produced from this modified Runx1lz allele contains the N-terminus of Runx1 through amino acid 242 (64 amino acids C-terminal to the DNA-binding Runt domain). Runx11z is a nonfunctional Runx1 allele [14]. The fusion of β-galactosidase to the DNA-binding domain of Runx1 could potentially interfere with endogenous core-binding factors by creating a protein that can bind DNA but not activate transcription. The fidelity of Runx1-β-galactosidase expression is supported by the observation that it correlates with the presence of Runx1 protein and mRNA in Runx1lz/+ embryos, as detected by immunohistochemistry and in situ hybridization [15, 16]. Here, mice heterozygous for this Runx1lz allele were used to isolate and characterize Runx1-expressing cells. We show that in the adult mouse, Runx1 was expressed in all BM hematopoietic stem cells (HSCs), in spleen colony-forming units (CFU-S), and in the majority of progenitor cells generating myeloid, erythroid and mixed colonies in culture. We also characterized Runx1 expression in more differentiated hematopoietic cells by flow cytometry and compared the BM and peripheral blood compositions of normal mice with those of mice heterozygous for two different mutated Runx1 alleles to determine which lineages are affected by reduced Runx1 dosage.

Materials and Methods

Mice

Two- to three-month-old male and female Runx1+/+, Runx1lz/+, and Runx1rd/+ mice on a mixed 129S3/SvImJ × C57BL/6J background [14, 17] were used throughout this study. Mice were housed according to institutional guidelines and animal procedures carried out in compliance with the standards for Humane Care and Use of Laboratory Animals.

Immunofluorescent Labeling and Flow Cytometry

Nucleated BM, spleen, and thymus cells were counted using a Neubauer hemocytometer (VWR Scientific; Mississauga, Ontario, Canada; http://www.vwrsp.com) and Türk's solution. Peripheral blood leukocyte counts were performed on whole blood using a Pentra 60 (ABI Diagnostics; Columbia, MD; http://www.abionline.com). Immunofluorescent labeling and flow cytometry were performed as described [14], with the modification that BM, spleen, and peripheral blood erythrocytes were lysed prior to labeling by a 2-minute incubation in ammonium chloride buffer (155 mM NH4Cl, 0.1 mM EDTA, 10 mM Tris-HCl, pH 7.5). All antibodies were obtained from Pharmingen (San Diego, CA; http://www.bdbiosciences.com/pharmingen) and were directed against c-kit (CD117; clone ACK45), Sca-1 (Ly-6A/E; D7), Mac-3 (M4/84), Gr-1 (Ly-6G; RB6-8C5), TER-119 (Ly-76; TER-119), Ly-6C (AL-21), BP-1 (Ly-51; 6C3), IgD (11-2c.2a), IgM (R6-60.2), NK1.1 (Ly-55; PK136), TCRβ (H57-597), TCRγδ (GL3), CD3 (145-2C11), CD4 (RM4-5), CD8 (53-6.7), CD11b (Mac-1; M1/70), CD11c (HL3), CD19 (1D3), CD24 (HSA; M1/69), CD25 (interleukin-2Rα; PC61), CD31 (PECAM-1; MEC 13.3), CD41 (MWReg30), CD43 (Ly-48; S7), CD44 (Pgp-1; IM7), B220 (CD45R; RA3-6B2), CD62P (RB40.34), and CD71 (C2). Fluorescein-di-β-D-galactopyranoside (FDG; Molecular Probes; Eugene, OR; http://www.probes.com) was used as a fluorescent substrate for β-galactosidase as described [18]. Briefly, cells were suspended in phosphate-buffered saline (PBS) supplemented with 10% fetal calf serum (FCS) and penicillin/streptomycin (P/S) and incubated for 1 minute precisely with an equal volume of FDG (2 mM stock in water) in a 37°C water bath. FDG loading was stopped by adding excess ice-cold PBS (with 10% FCS and P/S), and cells were kept on ice for 1 hour to recover prior to antibody labeling and throughout the entire labeling procedure. To determine the time span in which the fluorescent form of FDG could be detected, we measured its fluorescent product from 1 to 6 hours after loading at 1 hour intervals by flow cytometer. No changes in FDG signal were detected in this time span (data not shown), and cells were always analyzed within 2 to 3 hours of FDG loading. Wild-type hematopoietic cells were also loaded with FDG and served as negative controls for FDG signal. For antibody staining, appropriate isotype controls were used, as described previously [18]. A fluorescence-activated cell sorter, FACStar Plus (BD Biosciences; San Jose, CA; http://www.bdbiosciences.com), was used for cell sorting, and a CyAn (Dako-Cytomation; Carpinteria, CA; http://www.dakocytomation.com) and FACS Calibur (BD Biosciences) were used for analysis. Propidium iodide, Hoechst 33258, or TO-PRO3 (Molecular Probes) was used to exclude dead cells from analysis and cell sorting.

Analysis of Hematopoietic Activity

Long-term multilineage repopulation, CFU-S, and culture CFU (CFU-C) activity were determined as described [14, 18, 19] with minor modifications. BM cell populations were assayed for HSC activity by intravenous transfer into irradiated (900 rad, split dose) 3- to 6-month-old (129S3/SvImJ × C57BL/6J) F1 recipients. Recipients were analyzed for donor cell contribution at 1 month and 4–6 months post-transplantation by polymerase chain reaction of peripheral blood genomic DNA using primers for lacZ and myogenin sequences as described [18]. Multilineage engraftment was determined by flow cytometric analysis of BM cells as described [18]. For analysis of CFU-S11 activity, irradiated recipients (1,000 rad, split dose) were injected intravenously with sorted BM cells. Control mice were not injected. Spleens were harvested 11 days after transplantation and immersed in Tellyesniczky's solution (63% EtOH, 5% glacial HOAc, 2% formaldehyde in water). Colonies were counted by visual inspection. For analysis of CFU-C activity, sorted FDG-loaded BM cells and unsorted (not FDG loaded) BM cells were plated in 35-mm suspension culture dishes at 2 × 104 cells/dish in 1% methylcellulose in Iscove's medium (MethoCult 3231; StemCell Technologies; Vancouver, Canada; http://www.stemcell.com), supplemented with 50 ng/ml stem cell factor, 20 ng/ml interleukin (IL)-3, 25 ng/ml IL-6, 5 ng/ml GM-CSF, 2 ng/ml G-CSF, and 2 U/ml erythropoietin (all cytokines from R&D Systems; Minneapolis, MN; http://www.rndsystems.com). Cultures were grown at 37°C in 5% CO2, and hematopoietic colonies were counted after 11 days using an inverted microscope.

Western Blot Analysis

Single-cell suspensions were prepared from whole thymuses, lysed, and the nuclei collected and extracted in high salt buffer (20 mM HEPES, pH 7.5; 25% glycerol; 0.42 M NaCl; 3 mN MgCl2; 0.2 mM EDTA; 1 mM dithiothreitol; protease inhibitor cocktail [Sigma; St. Louis, MO; http://www.sigmaaldrich.com]). Nuclear proteins were separated on 12% SDS PAGE gels and transferred to nitrocellulose. Runx1 proteins were detected with mouse monoclonal antibody α 3.2.5 and enhanced chemiluminescence reagents (ECL; Amersham; Piscataway, NJ; http://www.apbiotech.com).

Results

Runx1-β-Galactosidase Fusion Protein Levels Are Similar to Wild-Type Runx1 Protein Levels

We utilized the β-galactosidase enzymatic reaction in cells from Runx1lz/+ mice to analyze Runx1 expression in various hematopoietic cell populations. β-galactosidase activity from the Runx1-β-galactosidase fusion protein should be a relatively sensitive indicator of Runx1 protein. However, it is possible that the stability of the Runx1-β-galactosidase fusion protein may differ from that of endogenous Runx1. To at least determine whether the steady-state levels of the Runx1-β-galactosidase fusion protein substantially differed from those of endogenous Runx1, we performed a Western blot analysis on thymic extracts prepared from Runx1lz/+ mice (Fig. 1) using a monoclonal antibody that recognizes the Runt domain. We found that the Runx1-β-galactosidase fusion protein did not accumulate to higher levels than the endogenous Runx1 protein.

Figure Figure 1..

Western blot analysis of thymic extracts.Thymocytes of Runx1lz/+and Runx1+/+mice were analyzed for endogenous and mutant Runx1 protein expression. Lanes 1, 2, and 6: thymic extracts of three Runx1lz/+mice; lanes 3–5: thymic extracts of three Runx1+/+mice. The position of Runx1 and Runx1-β-galactosidase proteins is indicated. Molecular weight markers in kilodaltons (kd) are listed on the left.

Adult BM-Derived Hematopoietic Stem and Progenitor Cells Express Runx1

Approximately 80% of cells in the adult Runx1lz/+ BM and 85% of the Linc-kit+ BM population, which is enriched for stem and progenitor cells, expressed Runx1 (Fig. 2A and 2B; Table 1). Runx1-expressing (Runx1+) and nonexpressing (Runx1) cells, isolated from total BM and from the c-kit+ and Linc-kit+ BM fractions, were assayed for the presence of HSCs. Upon intravenous transfer of 105 sorted cells into irradiated adult recipients, we found long-term multilineage high-level hematopoietic reconstitution, indicative of HSC activity, only in the Runx1+ populations (Table 1). The extent of the hematopoietic reconstitution was similar to that observed after transplantation of 105 unsorted BM cells (7 of 10 recipients reconstituted, 45%-73% donor-derived BM cells; not shown). No detectable donor cell reconstitution was observed after transfer of Runx1 cells. Thus, virtually all long-term repopulating HSCs express Runx1, beginning with their generation in ontogeny [18] and extending into adult life. We also determined that all progenitor cells capable of forming spleen colonies (CFU-S11) expressed Runx1 (Table 1). In addition, in vitro clonogenic progenitors such as CFU-granulocyte-erythrocyte-monocyte-megakaryocyte (CFU-GEMM), CFU-granulocyte-macrophage (GM), and BFU-E, were found almost exclusively among Runx1+ cells. The small number of CFU-C in the Runx1 fraction may be due to contamination with Runx1+ cells, or may reflect the presence of some CFU-C (sort purity was 98%).

Table Table 1.. BM hematopoietic stem and progenitor cells express Runx1
  1. a

    *Runx1+ and Runx1 BM cell populations isolated from Runx1lz/+ mice were tested for the presence of HSC activity by adoptive transfer into irradiated adult recipients as described in Materials and Methods. Recipients were considered positive for donor-derived reconstitution when they showed long-term, multilineage donor repopulation, with successful secondary transplantation. The range (%) of donor-derived cells in positive recipients is shown.

  2. b

    #84.8% ± 4% of Linc-kit+ cells fell within the Runx1+ sort gate and 11.0% ± 3% within the Runx1 gate.

  3. c

    §Irradiated adult recipients were transplanted with 5 × 104 sorted Runx1+ or Runx1 BM cells. Data are the mean ± SD of 10 mice per group. Residual endogenous hematopoietic activity (in a total of four irradiated, nontransplanted recipients) accounted for 0.3 ± 0.3 CFU-S11 per spleen.

  4. d

    Flow-sorted Runx1+ and Runx1 BM cells were assayed for their capacity to form hematopoietic colonies in methylcellulose. Data are the mean (range) of three independent experiments. The average relative distribution of CFU-GEMM over the two subsets was 96.6% ± 2.4% Runx1+ and 3.4% ± 2.4% Runx1; CFU-GM: 97.7% ± 2.1% Runx1+ and 2.3% ± 2.1% Runx1; BFU-E: 89.4% ± 14.5% Runx1+ and 10.6% ± 14.5% Runx1.

  5. e

    Abbreviations: BMC = bone marrow cells; n.a. = not applicable; SD = standard deviation.

Population% of BMC 
 (mean ± SD)ncells transferrednrecipients reconstituted% donor-derived BMCCFU-S11§(per 5×104cells)CFU-GEMMCFU-GMBFU-E
Runx1+81.6 ± 6.310516/1833.1 – 78.510.4 ± 4.34915738
      (32 – 78)(55 – 218)(15 – 57)
Runx112.5 ± 3.41050/15n.a.0.4 ± 0.41.62.52.7
      (0.4 – 2.5)(1.2 – 3.8)(0.6 – 5.7)
c-kit+ Runx1+4.4 ± 0.85 × 1035/531.3 – 55.0    
c-kit+ Runx11.1 ± 0.45 × 1030/4n.a.    
Lin c-kit+ Runx1+#1.4 ± 0.21046/722.2 - 61.9    
Lin c-kit+ Runx1−#0.2 ± 0.45 × 1030/5n.a.    
Figure Figure 2..

Flow cytometric analysis of Runx1 expression in BM cells.FDG was used as a fluorescent substrate forβ-galactosidase. In all experiments, Runx1+/+cells were used as a negative control for FDG loading (as shown in A, panel 1). Gates or markers for the FDG-negative cell populations in A-F were set based on wild-type FDG levels within the different lineage populations (as shown in B panel 2). Appropriate isotype control antibodies were used to determine background fluorescence. Dead cells were always excluded from analysis on the basis of propidium iodide, Hoechst 33258, or TO-PRO3 uptake. A) BM cells from Runx1lz/+and control +/+ mice were isolated and loaded with FDG. Sort gates are indicated, with representative percentages of cells within each gate. Reanalysis of sorted Runx1+and Runx1cells revealed cell purities of 95%-99%. Data are representative of 11 experiments. B) Analysis of Runx1 expression within the Linc-kit+BM population. BM cells were loaded with FDG and labeled with lineage (Mac-1, Gr-1, TER-119, B220, CD3, CD4, CD8) and c-kit antibodies. Sort gates are shown, with percentages of cells within each gate. Data are representative of four experiments. C) Changes in Runx1 expression during myeloid differentiation. Based on differential CD31 and Ly6-C expression [34], a population of CD31+Ly-6C+cells containing mainly (approximately 80%) myeloid progenitors and some erythroid progenitors, CD31Ly-6Cmedneutrophilic granulocytes (from band stage onwards), and CD31Ly-6Chimonocytes (from promonocyte stage onwards) can be distinguished. Histograms represent Runx1 (FDG) expression within each subset, and are representative of four experiments; average percentages are listed inTable6. The lower proportion of Runx1-expressing cells in the CD31+Ly-6C+population most likely resulted from the presence of erythroid cells that were largely Runx1(Figure1D). D) Changes in Runx1 expression during erythroid differentiation. BM cells were loaded with FDG and labeled with CD71 and TER-119 antibodies. In the spleen, this combination of antibodies has been shown to detect a maturation series of erythroid precursors with the order CD71+TER-119CD71hiTER-119+CD71medTER-119+CD71loTER-119+, representing proerythroblasts, basophilic erythroblasts, chromatophilic erythroblasts, and normoblasts, respectively [35]. In Runx1lz/+BM, similar phenotypic populations were identified. Morphological inspection of these BM populations after sort and Wright/Giemsa staining showed that they represent the same maturation sequence (data not shown). Histograms show Runx1 (FDG) expression within the four erythroid cell subsets and are representative of six experiments; seeTable6for average percentages. (E) Levels of Runx1 expression during T-cell maturation. Runx1lz/+CD4/8 DN, CD4/8 DP, CD4 SP, and CD8 SP thymocyte populations were analyzed for Runx1 expression. Representative FDG histograms of 1 out of 16 experiments are shown. SeeTable6for average Runx1 expression. F) Levels of Runx1 expression during B-cell maturation. Early B220+CD43+, and later B220+CD43stages of B-cell maturation [36] were analyzed for Runx1 expression. Representative FDG plots are from 1 experiment out of 14; seeTable6for average percentages of Runx1 expression. G) Schematic of percentages of Runx1+cells during T-cell maturation. Pie charts represent individual maturation stages; grey shaded areas represent the average percentages of Runx1-expressing cells within each cell population. The percentages of Runx1+cells (±SD) within the T-cell lineage are shown below the pies for thymic DN1 lymphoid progenitors (CD34844+25), DN2 pro-T (CD34844+25+), DN3 early pre-T (CD3484425+), DN4 late pre-T (CD3484425), DP (CD4+8+), CD4 SP (CD4+8), CD8 SP (CD48+), spleen CD4 SP (CD4+8), spleen CD8 SP (CD48+), thymic NK cell (CD348NK1.1+), andγδT cells (CD48TCRγδ+). No obvious changes compared with wild type were observed in the relative sizes of thymic NK andγδT cells, or in the relative sizes of DN1-4 subsets (data not shown). Data are from a total of 14 experiments. H) Percentages of Runx1+cells (±SD) within B-cell differentiation are shown for pre-pro-B (B220+CD43+HSABP1), pro-B (B220+CD43+HSA+BP1), large pre-B (B220+CD43+HSA+BP1+), small pre-B (B220+CD43IgM), immature B (B220+CD43IgM+IgD), mature B (B220+CD43IgM+IgD+), and spleen mature B (B220+CD43IgM+IgD+) cells. No difference in Runx1 expression was observed between the different spleen IgM+IgD+cells: IgMhighIgDlow(62% ± 6% Runx1+), IgMhighIgDhigh(65% ± 9%) and IgMlowIgDhigh(58% ± 6%) (not shown). Data are from a total of 10 experiments.

Table Table 6.. Runx1 expression and cellular composition of myeloid, erythroid, and lymphoid lineages from Runx1lz/+ and Runx1+/+ mice
  1. a

    Bone marrow cells and thymocytes (Runx1lz/+ and Runx1+/+) were labeled with cell type-specific antibodies. Isotype control antibodies were used as negative controls, and markers were set to include ≤0.5% of the negative cell population within the lineage+ gates. Runx1 expression was analyzed by FDG loading. FDG-loaded Runx1+/+ bone marrow cells and thymocytes served as negative control, and markers were set to include up to 1% of the negative cell population within the Runx1+ gate. Data are the mean ± SD of four to 16 samples. *p ≤ 0.05, #p = 0.005 compared with Runx1+/+ thymocytes (T-cell subsets), or BM cells (myeloid and erythroid subsets) (t-test). SD = standard deviation.

Cell typeMarkerRunx1+/+
  % lineage+cells (mean ± SD)% Runx1+cells within subset (mean ± SD)% lineage+cells (mean ± SD)
    • myeloid progenitor cellsCD31+ Ly-6C+8.9 ± 2*73 ± 411.3 ± 1
    • granulocytesCD31 Ly-6Cmed46 ± 288 ± 231.3 ± 2
    • monocytesCD31 Ly-6Chi5.8 ± 289 ± 74.3 ± 0.6
    • pro-erythroblastCD71+ TER-1191.1 ± 1.564 ± 170.7 ± 0.1
    • basophilic erythroblastCD71hi TER-119+6.4 ± 3.441 ± 57.0 ± 2.5
    • chromatoph erythroblastCD71med TER-119+3.3 ± 0.8#29 ± 46.8 ± 0.5
    • normoblastCD71lo TER-119+0.9 ± 0.516 ± 41.3 ± 0.3
    • DNCD4 CD89.1 ± 4*44 ± 116.8 ± 3
    • DPCD4+ CD8+76.8 ± 8*44 ± 1066.6 ± 8
    • CD4 SPCD4+ CD84.9 ± 2*76 ± 119.2 ± 1
    • CD8 SPCD4 CD8+1.5 ± 0.6*59 ± 113.4 ± 0.8
    • early B cellsB220+ CD43+12.6 ± 461 ± 914.3 ± 5
    • late B cellsB220+ CD4312.2 ± 446 ± 1313.8 ± 4

Decreased Cellularity in Hematopoietic Organs of Runx1lz/+ Mice

We consistently observed a 25% overall reduction in cellularity in BM and thymus from Runx1lz/+ mice compared with age- and sex-matched Runx1+/+ control mice (Table 2), while no significant decrease was observed in the spleen. In peripheral blood, no significant decrease was found either (Table 2), although we observed that four of seven mice analyzed had leukocyte counts between 1.1 and 1.8 × 106/ml, low counts that were not observed among the six wild-type mice analyzed.

Table Table 2.. Cellularity* of hematopoietic organs and peripheral blood in Runx1lz/+ and Runx1rd/+ mice
  1. a

    *Data are the mean ± SD of at least five age- and sex-matched mice per group. Numbers are per tissue (spleen and thymus), femur (bone marrow), or nucleated cells per ml peripheral blood. #p = 0.02, §p = 0.001, p = 0.0005 compared with wild type (t-test). SD = standard deviation.

TissueRunx1lz/+Runx1rd/+Wild type
Bone marrow1.8 × 107 (± 0.2)#1.6 × 107 (± 1.0)2.4 × 107 (± 0.3)
Thymus8.4 × 107 (± 0.6)§7.5 × 107 (± 1.4)§11 × 107 (± 0.7)
Spleen8.8 × 107 (± 1.7)7.8 × 107 (± 1.8)#9.6 × 107 (± 0.9)
Peripheral blood2.6 × 106 (± 1.7)3.0 × 106 (± 1.9)3.9 × 106 (± 2.3)

The Runx1-β-galactosidase fusion protein produced from the Runx1lz allele has the potential to bind DNA and dominantly inhibit Runx1-CBFβ function. To exclude the possibility that the consistent decrease in BM and thymus cellularity was specific to the Runx1lz allele, mice heterozygous for a nonfunctional Runx1 allele, in which an exon encoding a portion of the DNA-binding runt domain was deleted (Runx1rd/+ [17]), were also examined. A similar decrease in the cellularity of BM and thymus was seen in Runx1rd/+ and Runx1lz/+ mice (Table 2). In addition, the decrease in spleen cell count observed in the Runx1rd/+ mice versus wild-type mice was statistically significant. The difference in peripheral blood leukocyte counts between the Runx1rd/+ mice and wild-type mice was, like the difference between Runx1lz/+ and wild-type counts, not significant. However, we did find low leukocyte counts (1.1 and 1.6 × 106 nucleated cells/ml) in two of five Runx1rd/+ mice analyzed.

Small changes were seen in the cellular composition of BM and peripheral blood from Runx1lz/+ and Runx1rd/+ mice compared with BM and blood from wild-type mice (Tables 3 and 4). In BM there was a small but significant increase in the proportions of monocyte and granulocyte precursors (Mac-1 and Gr-1 cells), while for most other populations there were decreases in frequency (Table 3). In addition, we tested the effect of Runx1 gene dosage on early progenitor cells in functional assays. Similar frequencies of CFU-S, CFU-GEMM, CFU-GM, and BFU-E were obtained from both wild-type and Runx1lz/+ BM (Table 5). The adult stem cell compartment was not examined.

Table Table 3.. Changes in bone marrow composition detected in Runx1lz/+ and Runx1rd/+ mice
  1. a

    Bone marrow cells (Runx1lz/+, Runx1rd/+, and Runx1+/+) were labeled with cell type-specific antibodies. Isotype control antibodies were used as negative controls, and markers were set to include ≤0.5% of the negative cell population within the lineage-positive gates. Runx1 expression was analyzed by FDG loading of bone marrow cells. FDG-loaded Runx1+/+ bone marrow served as negative control, and markers were set to include up to 1% of the negative cell population within the Runx1+ gate. Data are the mean ± SD of at least six samples. The higher percentage of Runx1-expressing cells (70%) among the CD41+ bone marrow population compared with CD62P+ cells may be explained by the presence of hematopoietic progenitor cells as well as megakaryocytic cells [33].

  2. b

    *Number of lineage-positive cells as calculated per femur.

  3. c

    p < 0.05; **p < 0.01; #p < 0.005; p < 0.001; §p < 0.0001 compared with Runx1+/+ bone marrow (t-test). SD = standard deviation.

Cell TypeMarker
  % lineage+cells (mean ± SD)nlineage+cells*(×106)% Runx1+cells within subset (mean ± SD)% lineage+cells (mean ± SD)nlineage+cells*(×106)
Stem and progenitor cellsc-kit11 ± 21.8 ±0.3**73 ± 210 ± 42.4 ± 0.8
  (12.8 ± 1)(2.1 ± 0.1)   
 Sca-16.5 ± 21.2 ± 0.3#65 ± 108.3 ± 21.7 ± 0.3
  (8.6 ± 3)(1.4 ± 0.2)   
    • Monocytes/granulocytesMac-171.2 ± 9**12.7 ± 1.5#94 ± 1062.2 ± 714.6 ± 1.6
  (68.7 ± 3)**(11.9 ± 2)**   
    • GranulocytesGr-167.4 ± 7**12.0 ± 1.4#90 ± 859.5 ± 614.4 ± 1.8
  (65.1 ± 3)(10.5 ± 2.4)#   
    • MacrophagesMac-39.3 ± 51.7 ± 0.943 ± 311.1 ± 72.6 ± 1.5
  (10.2 ± 0.7)(1.8 ± 0.1)**   
    • MegakaryocytesCD411.6 ± 1**0.3 ± 0.2**70 ± 12.2 ± 10.5 ± 0.3
  (1.7 ± 0.2)**(0.3 ± 0.3)**   
 CD62P10.0 ± 3**1.8 ± 1.0§50 ± 213.3 ± 33.1 ± 0.6
  (10.7 ± 3)(1.7 ± 0.4)§   
    • Dendritic cellsCD11c16.6 ± 3§3.0 ± 0.5§58 ± 118.5 ± 36.6 ± 0.6
  (16.3 ± 3)(2.6 ± 0.4)§   
Erythroid cellsTER-11913.7 ± 42.3 ± 1.1#26 ± 319.5 ± 104.6 ± 2.3
  (15.6 ± 4)(2.5 ± 0.5)**   
B cellsB22024.0 ± 54.3 ± 1.8**43 ± 826.6 ± 46.3 ± 1.4
  (24.2 ±2)(3.9 ± 0.2)**   
 CD1914.2 ± 52.3 ± 1.0**50 ± 215.9 ± 43.7 ± 0.8
  (11.9 ± 2)(2.0 ± 0.2)**   
T cellsCD310.8 ± 62.0 ± 1.2**48 ± 913.3 ± 43.5 ± 0.8
  (9.8 ± 3)(1.5 ± 0.4)**   
 CD47.2 ± 1**1.1 ±0.2#60 ± 110.4 ± 52.4 ± 1.1
  (6.8 ± 2)**(1.0 ± 0.2)#   
 CD82.1 ± 1#0.3 ± 0.1#46 ± 13.0 ± 10.5 ± 0.2
  (2.0 ± 0.3)#(0.3 ± 0.1)#   
NK cellsNK1.13.9 ± 10.7 ± 0.274 ± 13.8 ± 20.9 ± 0.4
  (4.1 ± 0.6)(0.7 ± 0.1)   
Table Table 4.. Changes in peripheral blood composition detected in Runx1lz/+ and Runx1rd/+ mice
  1. a

    Peripheral blood leukocytes (Runx1lz/+, Runx1rd/+, and Runx1+/+) were labeled with cell type-specific antibodies. Isotype control antibodies were used as negative controls, and markers were set to include ≤0.5% of the negative cell population within the lineage+ gates. Runx1 expression was analyzed by FDG loading. FDG-loaded Runx1+/+ peripheral blood leukocytes served as negative control, and markers were set to include 1% of the negative cell population within the Runx1+ gate. Data are the mean ± SD of three to six samples. *p ≤ 0.05, §p < 0.01, #p < 0.005, p < 0.0005, compared with Runx1+/+ peripheral blood leukocytes (t-test). SD = standard deviation.

Cell TypeMarker
  % lineage+cells (mean ± SD)nlineage+cells(×106)% Runx1+cells within subset (mean ± SD)% lineage+cells (mean ± SD)nlineage+cells(×106)
MonocytesMac-1+ Gr-117.9 ± 6.90.5 ± 0.297 ± 111.6 ± 7.20.5 ± 0.3
  (11.2 ± 6.3)(0.3 ± 0.2)   
GranulocytesMac1+ Gr-1+9.5 ± 2.8*0.3 ± 0.199 ± 0.96.2 ± 2.10.2 ± 0.1
  (10.9 ± 2.6*)(0.3 ± 0.1)   
B cellsB22033.3 ± 9.30.9 ± 0.2#98 ± 139.0 ± 6.71.5 ± 0.3
  (34.8 ± 14.6)(1.0 ± 0.4#)   
T cellsCD47.6 ± 4.4§0.2 ± 0.184 ± 316.5 ± 4.50.6 ± 0.2
  (9.0 ± 5.0*)(0.3 ± 0.2*)   
 CD85.8 ± 3.1*0.2 ± 0.1#82 ± 611.2 ± 3.40.4 ± 0.1
  (10 ± 5.4)(0.3 ± 0.2)   
Table Table 5.. Comparison of CFU-S* and CFU-C# frequencies in Runx1+/+ and Runx1lz/+ BM
  1. a

    *Irradiated adult recipients were injected with 6 × 104Runx1+/+ or Runx1lz/+ BM cells. Data are per spleen (mean ± SD of six spleens per group).

  2. b

    #BM cells were isolated and plated in methylcellulose as described in Materials and Methods. Frequencies of CFU-C (mean ± SD of three independent experiments) are per 105 plated cells. SD = standard deviation.

PopulationCFU-SCFU-GEMMCFU-GMBFU-E
Runx1+/+ BM4.6 ± 1.246 ± 13153 ± 5139 ± 13
Runx1lz/+ BM6.0 ± 0.936 ± 22111 ± 5022 ± 9

Granulocytes and Monocytes Differ in Runx1 Expression Levels

Many of the oncogenic fusion genes involving Runx1 are associated with myeloid leukemias [1]. The defect is in part the failure of progenitor cells to differentiate properly, suggesting a role for Runx1 in myeloid differentiation, at least in the context of secondary mutations [1]. In BM myeloid differentiation, the large majority of myeloid blast cells, granulocytes from the band stage onwards, and monocytes from the promonocytic stage onwards expressed Runx1 (Fig. 2C). This is consistent with the reported broad expression of Runx1 mRNA and CBF family proteins in BM myeloid cells [10, 20] and the high frequency of Runx1+ cells detected among Mac-1+ and Gr-1+ cells in BM and peripheral blood (Tables 3 and 4). Notably, the level of Runx1 expression in BM granulocytes was approximately 4.5-fold lower than in monocytes (Fig. 2C). Runx1+ cells were also found among mature Mac3+ macrophages and CD11c+ dendritic cells in BM, albeit at lower frequencies (on average 43%-58%; Table 3). In the megakaryocytic lineage, on average, 50% of BM CD62P+ megakaryocytes (Table 3) and 59% ± 12% of peripheral blood CD41+ platelets (n = 4; not shown) expressed Runx1. CBFβ expression has also been reported in BM myeloid cells from CFU-GM through fully differentiated cells [13], supporting a role for Runx1-CBFβ throughout myeloid differentiation.

Runx1 Expression Decreases Upon Erythroid Maturation

Almost all BFU-E expressed Runx1 (Table 1), yet erythroid cells in BM are reported to express little or no RUNX1 protein and mRNA [10, 11] or other proteins of the CBF family [13, 20]. Indeed, only 26% ± 3% of TER-119+ cells were found to express Runx1 (Table 3). Examination of Runx1 expression during erythroid maturation showed that while, on average, 64% of proerythroblasts still expressed Runx1, this percentage decreased upon maturation to only 16% ± 4% of normoblasts (Fig. 2D, Table 6); the levels of Runx1 expression in individual cells also decreased (Fig. 2D). A similar decrease in CBFβ expression was reported [13]. Thus, Runx1-CBFβ could be involved in the initial activation of erythroid-specific genes at the CFU-GEMM or BFU-E stages [21], but is probably not required for terminal erythroid maturation [11, 22].

Fewer Mature Than Immature Lymphoid Cells Express Runx1

Runx1 has been shown to play a role at distinct stages in T-cell development [12, 23, 24], and Runx1 expression has been detected in the thymic cortex, the major CD4/CD8-defined thymocyte populations, and spleen T cells [59, 12, 23, 25]. In addition, a translocation involving RUNX1 has recently been reported in acute T-lymphoblastic leukemia, and proviral insertion studies indicated Runx1 as a dominant oncogene in T-cell lymphoma [26, 27]. CBFβ protein expression was reported in all stages of T-cell differentiation [13].

T-cell development in the thymus is characterized by the maturation of CD4 and CD8 single-positive (SP) cells from a lymphoid precursor population. Hayashi et al. [9] showed an effect of decreased Runx1 dosage in thymocyte maturation characterized by reduced numbers of CD4 and CD8 SP T cells and a relative increase in the percentage of CD4/CD8 double-positive (DP) T cells. In accordance with this, we observed similar changes in CD4/CD8 populations in thymocytes and reduced numbers of CD4 and CD8 SP cells in BM and peripheral blood from Runx1lz/+ mice (Tables 3, 4, and 6). We examined Runx1 expression within the distinct CD4/CD8 thymocyte populations. On average, 44% ± 11% of double negative (DN) cells expressed Runx1 (Fig. 2E; Table 6). The majority of cells in the first three DN stages expressed Runx1 (Fig. 2G), with no observed differences in levels of Runx1 expression between the three subsets (data not shown). In the fourth DN stage, the percentage of Runx1-expressing cells decreased dramatically (Figure 2G). The percentage of Runx1+ cells increases at later stages of T-cell differentiation (Fig. 2E and 2G; Tables 3, 4, and 6).

A role for Runx1 in B lymphopoiesis is suggested by the association of the t(12;21) TEL-Runx1 fusion protein with pediatric acute B-lymphocytic leukemia [1]. We found Runx1 expressed at all stages of differentiation (Fig. 2F and 2H; Table 6) and in peripheral blood B cells (Table 4). During B-cell differentiation in the BM, on average, 48%-68% of the cells are Runx1+ (Fig. 2F and 2H; Table 6), with no obvious differences in the levels of Runx1 expression between immature and more mature B-cell populations (Fig. 2F, and data not shown).

Discussion

A crucial role for Runx1 in the generation of the definitive hematopoietic system in ontogeny has been demonstrated [14, 17, 2829]. Here we show that in adult mouse hematopoiesis, Runx1 is expressed in virtually all hematopoietic stem and progenitor cells and continues to be expressed in most cells differentiating into the monocyte/granulocyte lineages. In the lymphoid lineages, Runx1 is expressed at all maturation stages, with variable proportions of Runx1+ cells at the different stages. In the erythroid lineage, Runx1 expression decreases after the proerythroblast stage.

The basis for the varying percentages of Runx1+ and Runx1 cells in the lymphoid subpopulations is currently unclear. It may reflect differences in Runx1 expression in cycling versus resting cells. Another possibility is that Runx1 expression in lymphoid cells is stochastic, such that only a certain percentage of cells are at any time Runx1+. Or the various populations we examined may be heterogeneous and consist of specific subpopulations of Runx1+ or Runx1 cells that could be further segregated based on the expression of additional markers. For example, it was recently shown that T helper (Th) 1 and Th2 cells differ in Runx1 expression [24]. Since CBFβ is expressed uniformly throughout T-cell development [13], it will be of interest to examine whether Runx1 cells express Runx2 and/or Runx3, the other two core-binding factor family members for which important roles in T-cell differentiation have been reported [12, 23, 30]. The highest proportion of Runx1 cells in the lymphoid lineages was observed in the last DN stage. It is possible that this may reflect the recently reported function of Runx1 in CD4 repression [12, 23], in that the decrease in Runx1 expression might allow the cells to proceed to the CD4+CD8+ stage. However, substantiation of this notion awaits more detailed studies.

We showed by Western blot analysis that the Runx1-β-galactosidase fusion protein is expressed at similar or slightly lower steady-state levels than endogenous Runx1 protein. It remains possible that the Runx1-β-galactosidase protein is degraded at a different rate than endogenous Runx1. Human RUNX1 is degraded via the ubiquitin-proteasome pathway, and heterodimerization of RUNX1 with CBFβ, through the Runt domain, protects RUNX1 from degradation [31]. At least five of the lysine residues targeted by ubiquitin-mediated degradation are conserved in mouse Runx1 [31] and are maintained in Runx1-β-galactosidase. Furthermore, Runx1-β-galactosidase still contains the Runt domain [14], and CBFβ was reported to be expressed in all hematopoietic cell types in which we detected Runx1 [13]. Thus, two important aspects influencing Runx1 protein stability are still available to the fusion protein. Uncertainties about protein turnover would apply to any of the currently available methodologies used to study gene expression with fluorescent proteins. For example, expression of β-galactosidase or green fluorescent protein from the Runx1 locus via an internal ribosome entry site rather than as a fusion protein would bypass the intricate translational control to which Runx1 mRNAs are normally subjected [32], and the stability of the marker protein would not necessarily mimic that of endogenous Runx1. Comparison of enzymatic or fluorescent activity with Runx1 protein levels is also difficult due to differences in the sensitivity of detection. The approach we have taken would overestimate the population of Runx1+ cells if the fusion protein turns over less rapidly than endogenous Runx1. The observation that all HSCs and virtually all committed progenitors fall into the Runx1+ population suggests it is unlikely that we are underestimating the percentage of cells in which Runx1 is expressed. The heterogeneity in levels of Runx1 expression detected by flow cytometry in total BM Runx1+ cells could not be unambiguously related to one or more specific cell types or differentiation stage(s). Instead, the variation was seen in most Runx1+ cell populations.

BM and thymus of Runx1lz/+ mice consistently showed a decreased cellularity. In addition, small but significant changes in the frequencies of the T-cell and monocyte/granulocyte lineages were observed. Decreased cellularity of hematopoietic tissues and changes in their cellular composition were also seen in Runx1rd/+ mice, which are heterozygous for the Runx1rd allele from which a DNA-binding defective, unstable, or no protein would be produced (Tables 3 and 4), indicating that the effects are not specific to the Runx1lz allele. Similar effects on the frequencies of cells in the T-cell or monocyte/granulocyte lineages were found in Runx1lz/+ and Runx1rd/+ mice. In the T-cell lineage, reduced frequencies of mature SP CD4 and CD8 T cells were observed, as reported previously [9]. In contrast, in the myeloid lineage, frequencies of monocytes (BM) and granulocytic cells (BM and peripheral blood) were elevated. As the frequency of Runx1lz/+ CFU-GM was comparable to wild type, the relative increase in myeloid cells cannot be explained by an increase in progenitor cells. Indeed, an increase in frequency was observed only in the terminal stages of BM monocytic (CD31-Ly-6Chi) and granulocytic (CD31-Ly-6Cmed) maturation, but not in the stage before (CD31+Ly-6C+; Table 6 and Fig. 2C). Elucidation of the mechanism(s) behind the observed perturbations in hematopoiesis awaits future studies. The expression of Runx1 in functional HSCs and its continued expression in adult myeloid and lymphoid lineage cells support a role for Runx1 in HSC biology and terminal differentiation of the myeloid and lymphoid lineages. Studies using Runx1 conditional knock-out mice will help to determine at precisely what stage(s) Runx1 is required, and to define whether Runx1 acts as a transcriptional repressor and/or activator in those cells.

Acknowledgements

We thank Dr. Alice Givan, Gary Ward, Ann Atzberger, and Christiane Kuhl for their help with flow cytometry, and Drs. Rudi Hendriks, Tariq Enver, Catherine Porcher, and Paresh Vyas for helpful discussions. T.E.N. and C.J.M. were supported by T32GM08704 from the NIH/GM and N.A.S. by Public Health Service grant R01CA58343. M.F.T.R.d.B. was supported by a fellowship from the Dutch Cancer Society. Flow cytometry at Dartmouth Medical School was done in The Herbert C. Englert Cell Analysis Laboratory, established by a grant from the Fannie E. Rippel Foundation and supported in part by the Core Grant of the Norris Cotton Cancer Center (CA 23108).

Ancillary