Enforced Expression of PU.1 Rescues Osteoclastogenesis from Embryonic Stem Cells Lacking Tal-1


  • Motokazu Tsuneto Ph.D.,

    Corresponding author
    1. Division of Immunology, Department of Molecular and Cellular Biology, School of Life Science, Faculty of Medicine, Tottori University, Yonago, Tottori, Japan
    • Division of Immunology, Department of Molecular and Cellular Biology, School of Life Science, Faculty of Medicine, Tottori University, 86 Nishi-Machi, Yonago, Tottori, 683-8503, Japan. Telephone: 81-859-34-8270; Fax: 81-859-34-8272
    Search for more papers by this author
  • Akira Tominaga,

    1. Department of Molecular and Cellular Biology, Kochi Medical School, and Laboratory of Human Health and Medical Science, Graduate School of Kuroshio Science, Nankoku, Kochi, Japan
    Search for more papers by this author
  • Hidetoshi Yamazaki,

    1. Division of Immunology, Department of Molecular and Cellular Biology, School of Life Science, Faculty of Medicine, Tottori University, Yonago, Tottori, Japan
    2. Division of Regenerative Medicine and Therapeutics, Department of Genetic Medicine and Regenerative Therapeutics, Institute of Regenerative Medicine and Biofunction, Tottori University Graduate School of Medical Science, Yonago, Tottori, Japan
    Search for more papers by this author
  • Miya Yoshino,

    1. Division of Immunology, Department of Molecular and Cellular Biology, School of Life Science, Faculty of Medicine, Tottori University, Yonago, Tottori, Japan
    Search for more papers by this author
  • Stuart H. Orkin,

    1. Department of Pediatric Oncology, Children's Hospital and the Dana Farber Cancer Institute, Harvard Medical School and Howard Hughes Medical Institute, Boston, Massachusetts, USA
    Search for more papers by this author
  • Shin-Ichi Hayashi

    1. Division of Immunology, Department of Molecular and Cellular Biology, School of Life Science, Faculty of Medicine, Tottori University, Yonago, Tottori, Japan
    Search for more papers by this author


Transcription factor T-cell acute lymphocytic leukemia 1 (Tal-1) is essential for the specification of hematopoietic development. Mice lacking Tal1 fail to generate any hematopoietic precursors. Using our co-culture system with stromal cells, we demonstrate that enforced expression of the transcription factor PU.1 under tetracycline control in Tal1-null embryonic stem (ES) cells rescues the development of osteoclasts and macrophage–like phagocytes. It was low efficiency compared with wild–type ES cells; other hematopoietic lineage cells of granulocytes, B cells, mast cells, megakaryocytes, and erythroid cells were not generated. Osteoclasts developed in this culture were multinucleated and competent for bone resorption. Their development depended on macrophage colony-stimulating factor and receptor activator of nuclear factor κB ligand. The majority of cells with the potential to differentiate into osteoclasts expressed fetal liver kinase 1 (Flk-1) and could be isolated using anti–Flk-1 antibody. These results suggest that the expression of PU.1 is a critical event for osteoclastogenesis and that Tal-1 may lie upstream of PU.1 in a regulatory hierarchy during osteoclastogenesis.


Hematopoiesis can be viewed as a hierarchy with hematopoietic stem cells (HSCs) at the top and progenitors and their descendents below. Tal1 gene knockout (Tal1−/−) mice succumb during embryogenesis due to a complete failure to produce blood cells [1, 2]. The product of the Tal1 gene is believed to be essential for the initiation of hematopoietic development and the formation of HSCs. Consistent with the absence of hematopoietic cells and their descendants, transcription factor PU.1 (gene symbol, Sfpi1) is not detected in Tal1−/− embryonic stem (ES) cells induced to form hematopoietic cells in vitro [3] (in this report). PU.1 controls development into myeloid and B-lymphoid cell lineages by regulating, in part, the receptors for macrophage colony-stimulating factor (M-CSF), Fms [4], and interleukin-7 (IL-7R)α [5], respectively. Myelopoiesis, B lymphopoiesis, and osteoclastogenesis are impaired in PU.1-deficient mice [6, 7]. Therefore, although it is not clear whether the expression of PU.1 is directly regulated by Tal-1, we wondered whether deficiency of Tal-1 might lead to failure to induce PU.1, resulting in the absence of some (or all) PU.1-dependent lineages.

Osteoclasts resorb bone matrices. They are ultimately derived from HSCs. Osteoclasts are distinguished by their multinuclearity and expression of tartrate-resistant acid phosphatase (TRAP) [810]. Osteoblasts and stromal cells support osteoclastogenesis by supplying essential factors, such as M-CSF [11] and receptor activator of nuclear factor κB ligand (RANKL) [1214]. Previously we developed a culture system for osteoclastogenesis in which ES cells are co-cultured with stromal cells [15]. This system enables access to the entire program of osteoclastogenesis from undifferentiated ES cells to mature functional osteoclasts.

To assess the potential function of PU.1 downstream of Tal-1 in osteoclastogenesis, we have expressed exogenous PU.1 in Tal1−/− ES cells using the tetracycline (Tc)-off system [16, 17]. Surprisingly, we observe that enforced PU.1 expression induced osteoclast differentiation, even though the efficiency was lower than in wild-type ES cells. In this context, osteoclast development depended on M-CSF and RANKL, as in normal osteoclastogenesis. These findings suggest that PU.1 is a critical transcription factor for osteoclastogenesis; they are consistent with a role for PU.1 downstream of Tal-1 in a pathway culminating in osteoclastogenesis.

Materials and Methods

Cell Lines

A bone marrow–derived stromal cell line, ST2 [18], was maintained in RPMI-1640 (Roswell Park Memorial Institute: Gibco-Invitrogen Corp., Grand Island, NY, http://www.invitrogen.com), supplemented with 5×10−5 M 2-mercaptoethanol (2ME: Wako Pure Chemical Industries, Osaka, Japan; http://www.wako-chem.co.jp/english) and 5% fetal bovine serum (FBS: JRH Biosciences, Lenexa, KS; http://www.jrhbio.com). The OP9 stromal cell line [19] was cultured in minimum essential medium alpha (αMEM: Gibco-Invitrogen) supplemented with 20% FBS. ES cell lines J1 and Tal1−/− J1 [2], and Sfp/1 PU1-transfected Tal1−/− J1 ES cell lines were maintained in Dulbecco's modified essential medium (DMEM: Gibco-Invitrogen) supplemented with 10% knockout serum (Gibco-Invitrogen), 1% heat-inactivated FBS, 10−4 M 2ME, 1× nonessential amino acids (Gibco-Invitrogen), 2 mM L-glutamine (Gibco-Invitrogen), and leukemia inhibitory factor (LIF) equivalent to 1,000 U/ml on 0.1% gelatin-coated culture dishes. The RAW 264.7 macrophage cell line was maintained in αMEM supplemented with 10% FBS (JRH).


Tal1−/− J1 ES cells were first transfected with Tc-regulated transactivator (tTA) driven by the CAG promoter [20] of the Tc-responsive promoter (CMV*-1)-puror (a gift from Dr. H. Niwa, RIKEN Kobe, Japan). These cells were cultured in the presence of 1 μg/ml puromycin and LIF for 8 days, and growing colonies were recovered. After Tc was added to the culture medium, clones sensitive to puromycin were chosen. These clones were secondarily transfected with a CMV*-1-ligated mouse Sfpi1 cDNA sequence, followed by an internal ribosomal entry site (IRES) and green fluorescent protein (GFP)(CMV*-1-PU.1-IRES-EGFP, Fig. 1A).PSV2-bsrwas simultaneously transfected. These cells were cultured with 1 μg/ml Tc and 3 μg/ml blastocidin S hydrochloride (blastocidin: Kaken Pharmaceutical Co., Tokyo, Japan; http://www.nni.nikkei.co.jp) for an initial 4 days, and during the following period, 0.5 μg/ml blastocidin was added to cultures. On day 12, colonies were picked up, and clones whose expression of GFP was regulated by Tc were chosen.

Figure Figure 1..

Establishment of Tc-controlled PU.1-expressing Tal1−/−ES cell lines. (A): Constructs transfected into Tal1−/− ES cells (see Materials and Methods). (B, C): The expression of PU.1 was evaluated by monitoring the fluorescence of GFP. ES cells were cultured in the presence of LIF, and flow cytometric analysis was performed 1 or 2 days after Tc was withdrawn (B). Microscopic analysis of phase contrast field and fluorescent microscopic fields of TUNE ES cells cultured with TC (Tc+) or without Tc (Tc−) TC for 2 days in upper and lower rows, respectively (C). (D): PU.1 protein was detected in cell lysates prepared on day 2 or day 3 from TUNE ES cells cultured on OP9 cells after Tc was withdrawn starting on day 0, and from a macrophage cell line, RAW264.7 (RAW). Cell lysates from TUNE ES cells cultured in the presence of Tc throughout the culture period (Tc+) or from OP9 cells did not contain detectable PU.1 protein. (E): Schematic diagram of culture systems in this study.

Differentiation of ES Cells

Undifferentiated ES cells were inoculated at 104 cells per well in six-well plates (Corning, NY; http://www.corning.com) on pre-seeded OP9 cells and cultivated in αMEM supplemented with 20% FBS (Thermo Trace, Melbourne, Australia; http://www.thermotrace.com.au/). On day 5, the cells were harvested and re-seeded at 105 cells per well in six-well plates onto confluent OP9 layers [19] in αMEM containing 20% FBS. Five days later, cells were harvested and osteoclasts were induced on ST2 stromal cells at 103 cells per 24-well plate (Corning) in αMEM/10% FBS (JRH) supplemented with 10−8 M 1α,25(OH)2D3 (Biomol Research Laboratories, Plymouth Meeting, PA; http://www.biomol.com) and 10−7 M dexamethasone (Dex: Sigma Chemical Corp., St. Louis, MO; http://www.sigma-aldrich.com). Six days later, TRAP staining was performed and TRAP+ cells were counted under a microscope. These cultures are referred to as the osteoclast cultures (Fig.1E) [21]. Inthepit-formation assay, cells derived from ES cells were co-cultured with ST2 cells in the presence of 10−8 M 1α,25(OH)2D3, 10–7 M Dex, 10 μg/ml human M-CSF (provided by Dr. M. Takahashi, Otsuka Pharmaceutical, Tokyo), and 25 ng/ml human soluble RANKL (PeproTech, Rocky Hill, NJ; http://www.peprotech.com) on dentine slices (a gift from Dr. N. Udagawa, Matsumoto Dental University, Shiojiri, Japan).

Western Blot Analysis

Proteins were extracted from cultured cells by lysing the cells with lysis buffer containing EDTA and Triton X-100. Each sample (300 μg of protein) was separated by SDS-polyacrylamide gel electrophoresis (SDS-PAGE) and transferred to a nitrocellulose membrane (Hybond ECL: Amersham Pharmacia Biotech, Piscataway, NJ; http://www.amershambiosciences.com). After the membrane was blocked with 5% skim milk (DIFCO Laboratories, Detroit, MI; http://www.vgdusa.com/DIFCO.htm), it was incubated with rabbit anti-mouse PU.1 antibody. PU.1 protein was visualized by using horseradish peroxidase (HRP)–conjugated goat anti-rabbit immunoglobulin-G (IgG) (Amersham Pharmacia Biotech) and the enhanced chemiluminescence (ECL) detection system (Amersham Pharmacia Biotech). Protein extract from macrophage cell line RAW264.7 was used as positive control.

TRAP Staining

Cultured cells were fixed with 10% formaldehyde (Wako Pure Chemical Industries) for 10 minutes at room temperature and with ethanol/acetone (50:50 v/v: Wako) for 1 minute. After the cells were washed with PBS, they were stained with fast red violet LB-salt (Sigma) mixed with TRAP solution containing 59.3 M of sodium tartrate (Wako), 165.7 M sodium acetate (Wako), and 0.56 mg/ml naphthol AS-MX phosphate (Sigma) for 5 minutes at room temperature. Red-stained cells were visualized under a microscope and counted as TRAP+ cells.

Reverse Transcription–Polymerase Chain Reaction (RT-PCR)

Total RNA was purified using ISOGEN (Nippon Gene, Toyama, Japan; www.nippongene.jp) and used as the template to synthesize cDNA using RevaTra Ace (Toyobo, Osaka,Japan;http://www.toyobo.co.jp/e) primed with oligo dT from 1 μg of total RNA. Gene expression was analyzed by PCR using the following primers:


Both endogenous and exogenous Sfpl1:5′-GGA GAC AGG CAG CAA GAA AA-3′ and 5′-GCG ACG GGT TAA TGC TAT GG-3′).

Only endogenous Sfpl1: 5′-TTG ATC CCC ACC GAA GCA GG-3′ and 5′-ATG TGG CGA TAG AGC TGC TG-3

Hbb (b-hemoglobin): 5′-CAC AAC CCC AGA AAC AGA CA-3′ and 5′-CTG ACA GAT GCT CTC TTG GG-3′

Hbb-bh1 (ζ-hemoglobin): 5′-GCT CAG GCC GAG CCC ATT GG-3′ and 5′-TAG CGG TAC TTC TCA GTC AG-3′



Cpa3 (mast cell carboxypeptidase A precursors; mMC-CPA): 5′-CCG TGT GAA ACT CCA GAA TG-3′ and 5′-GTG CAT GAA TGC CAC AGT CC-3′

Ppbp (platelet basic protein; PBP): 5′-ATG GGC TTC AGA CTC AGA CC-3′ and 5′-GAC GAT TCT CTT GAC GCC AG-3′



PCR was performed under the following conditions: an initial cycle consisting of 94°C for 4 minutes, annealing at 60°C for 3 minutes, and 72°C for 1 minute, followed by 34 cycles or 44 cycles of 94°C for 1 minute, annealing at 60°C for 1 minute, and 72°C for 1 minute. cDNA prepared from cells expressing each gene tested was used as a positive control (Posi). Distilled water (DW) was used for negative control.

Detection of Phagocytes

After ES cells were cultured on OP9 layers for 5 days in αMEM supplemented with 20% FBS (Thermo Trace), phagocytes were induced on pre-seeded ST2 layers for 6 days inαMEM supplemented with 10% FBS (Thermo Trace) and M-CSF. Cells were incubated with fluoresbrites microspheres (Poly-Science, Niles, IL; http://www.polyscience.com) for 2 hours at 37°C. After washing, these cells were harvested by trypsinization. They were re-seeded into 100-mm culture dishes (Corning) to easily distinguish phagocytes from stromal cells. Under fluorescent microscopy, the number of phagocytes that had more than 20 fluorescent beads was counted. To observe appearances of phagocytes, the cells were recovered, cytospun, and stained with May-Grunwald-Giemsa solution.

Immunochemical Staining

Phagocytes were cultivated on cover glasses (Matsunami Glass Ind., Osaka, Japan; http://www.matsunami-glass.co.jp/e) and were incubated with latex beads (Sigma) for 2 hours at 37°C. After washing, cells were fixed with methanol for 15 minutes and then incubated with Bloking Ace (Snow Brand Milk Products, Co. Ltd., Tokyo, japan; http://www.snowbrand.co.jp) for 20 minutes at 4°C. Macrosialin was detected with rat anti-mouse macrosialin antibody (FA11; Serotec, Raleigh, NC; http://www.serotec.com), biotinylated-goat anti-rat IgG antibody (KPL, Kirkegaard & Perry Laboratories, Gaithersburg, MD; http://www.kpl.com), streptavidin-peroxidase (KPL), and DAB reagent set (KPL).


Establishment of PU.1-Expressing Tal1−/− ES Cell Lines under the Tc-off System

To provide a system for studying the regulated expression of PU.1, cDNA was placed under the control of the Tc-off system and introduced into Tal1−/− ES cells (Fig. 1A). Three independent ES clones (designated TUNE-1 to TUNE-3) were thereby established. Expression of the exogenous PU.1 wasinhibitedcompletelyby100ng/mlTc,andexpressionwas induced by withdrawal of Tc from the culture. The expression of PU.1 was evaluated by monitoring the fluorescence of GFP (Fig. 1B, C). The expression of exogenous PU.1 reached a maximum level 1 day after Tc withdrawal (Fig. 1B). All three clones behaved similarly; therefore, representative data from TUNE-1 ES cells are shown unless otherwise indicated.

We used a stepwise culture system to induce osteoclasts from ES cells (Fig. 1E, osteoclast culture). PU.1 protein was detected on days 2 and 3 in PU.1-expressing (PU.1-on) TUNE ES cells in the osteoclast cultures but not in TUNE ES cells cultured in the presence of Tc (PU.1-off) or in OP9 cells (Fig. 1D).

Osteoclastogenesis from PU.1-Expressing TUNE ES Cells

Myelopoiesis and osteoclastogenesis are impaired in PU.1-null mice. To determine if exogenous PU.1 could rescue osteoclastogenesis in TUNE ES cells, TUNE ES cells were cultivated in osteoclast cultures. Multinucleated TRAP+ cells displaying more than six nuclei were induced from PU.1-on TUNE ES cells following Tc withdrawal on day 2 or 3 of culture. Although cells derived from TUNE ES cells existed, no TRAP+ cells were generated when Tc was present (Fig. 2A). This is consistent with our previous report that no TRAP+ cells were induced from Tal1−/− ES cells [22]. PU.1 expression was restricted in GFP+ fraction (data not shown), and TRAP+ cells were mainly induced from GFP+ fraction. Non-specific toxicity of Tc could not account for these findings, as the number of TRAP+ cells derived from parental J1 ES cells was not affected by the presence of Tc.

Figure Figure 2..

Enforced expression of PU.1-rescued osteoclastogenesis from Tal1−/− ES cells. TUNE ES cells were cultured on OP9 cells for 10 days. Tc was withdrawn from day 2 (Tc−). The cells harvested on day 10 were induced to form osteoclasts on ST2. (A): Osteoclasts were induced in the schedule as described in Figure 1E (osteoclast culture). TRAP–positive cells were detected as red cells by TRAP staining (upper row). Osteoclasts were induced on dentine slices by co-culturing with ST2 in the presence of 1α,25(OH)2D3, Dex, M-CSF, and RANKL for a further 29 days. Pits are indicated by an arrowhead (lower row). Neither osteoclast nor pits was observed in the presence of Tc (Tc+). (B): After ES cells were cultivated on OP9 cells for 10 days, the cells harvested from cultures of parental wild-type J1 ES cells, or TUNE ES cells continuously treated with Tc (Tc) or from which Tc was withdrawn on day 2(−) were co-cultured on ST2 layers in the presence of 1α,25(OH)2D3 and Dex for 6 days. TUNE ES cells from which Tc had been withdrawn were cultured with an anti-Fms antagonistic antibody (Anti-Fms) or OPG during the final 6 days. Each column represents the mean ± SD of triplicate cultures.

To determine whether the TRAP+ cells were functional osteoclasts, a pit-formation assay was performed. On day 10 of the osteoclast culture, harvested cells were induced to form osteoclasts on dentine slices by co-culturing with ST2 cells supplemented with 1α, 25(OH)2D3, Dex, human M-CSF, and human RANKL. On day 39, pit formation was observed in cultures of PU.1-on TUNE ES cells, whereas no pits were formed under the PU.1-off condition (Fig. 2A). These results indicate that mature functional osteoclasts were induced from Tal1−/− ES cells following expression of PU.1.

M-CSF and RANKL are known to be critical for osteoclastogenesis. To investigate the requirement for these factors, an antagonistic anti-Fms antibody or a decoy receptor of RANKL, osteoprotegerin (OPG) [23] was added to the final phase of the osteoclast cultures on ST2 cells. Both factors completely inhibited osteoclastogenesis from PU.1-on TUNE ES cells (Fig. 2B). These results demonstrate that M-CSF and RANKL are also essential for the development of osteoclasts derived from PU.1-on TUNE ES cells.

Macrophage-Like Phagocytes Were Induced from PU.1-on TUNE Cells

PU.1 is an essential transcription factor for differentiation of macrophages. Since osteoclast precursors are thought to be derived from monocytelineage cells, it may be possible to induce monocyte and macrophage lineages by enforced expression of PU.1. Therefore, PU.1-on TUNE cells were cultivated on OP9 stromal layers for 5 days, then co-cultured with ST2 stromal cells supplemented with M-CSF for a further 6 days. Finally, these cells were incubated with fluorescent microspheres. Phagocytes that had more than 20 microspheres were counted under the fluorescent microscopy.

The numbers of phagocytes induced from TUNE-2 cells was one-fortieth compared with that from J1 wild type ES cells. Few cells with microspheres were observed in the cultures of Tal1−/− J1 ES cells and PU.1-off TUNE-2 cells (Fig. 3A). These cells were cytospun and stained with May-Grunwald-Giemsa solution to observe their appearances. The bead-carrying cells from J1 ES and PU.1-on TUNE-2 cells looked like monocytes or macrophages (Fig. 3B). In contrast, almost all cells with microspheres in Tal1−/− ES and PU.1-off TUNE-2 cell cultures were the feeder stromal cells (Fig. 3B).

Figure Figure 3..

Macrophage-like phagocytes were induced from PU.1-on TUNE-2 cells. After ES cells were cultured on OP9 layers for 5 days, phagocytes were induced on ST2 layers supplemented with M-CSF for 6 days. Phagocytes with fluorescent beads (B) or nonfluorescent latex beads (C) were detected under microscopy. (A): The number of phagocytes was counted under fluorescent microscopy. J1: wild-type ES cells; Tal1−/−: Tal-1-deficient J1 ES cells. (B): Cytospun cells were stained with May-Grunwald-Giemsa solution (upper row). Macrophage-like cells (arrowheads: upper row) with fluorescent microbeads (lower row). Tc−: cells cultured without Tc; Tc+: cells cultured with Tc. (C): Cells were stained with anti-macrosialin antibody. Macrosialin-positive cells were observed as brown. Isotype control: instead of anti-macrosialin antibody, anti-B220 antibody was used as the first antibody.

To further clarify their phenotype, phagocytes were stained with the antibody directed to macrosialin, which is macrophage-restricted antigen [24]. Macrosialin was detected in phagocytes with latex beads from cultures of both wild-type J1 and PU.1-on TUNE cells, while no macrosialin-positive cells were detected in cultures of Tal1−/−deficient ES cells. This is accordance with observation that no hematopoiesis occurs from Tal1−/− ES. These results suggest that monocyte- and macrophage-like phagocytes were also induced from Tal-1–deficient ES cells by enforced expression of PU.1.

Erythrocytes, Mast Cells, and Megakaryocytes Were Not Induced from TUNE-1 ES Cells

To assess whether other hematopoietic lineages might be rescued by the expression of PU.1, transcripts for several lineage-associated genes were examined by RT-PCR. ES cells were cultured on OP9 cell layers, and on day 5, cDNAs were synthesized from total RNA of cultured cells. Exogenous PU.1, but not endogenous PU.1, was expressed. Lineage-related genes (β-globin and ζ-globin for erythrocytes, mMC-CPA for mast cells, and platelet basic protein [PBP] for megakaryocytes) were analyzed. Transcripts for these markers were detected in cultures of wild-type J1 ES cells; none, however, were observed in PU.1-on TUNE ES cells (Fig. 4). These data indicate that exogenous PU.1 fails to rescue the development of mast cells, erythrocytes, and megakaryocytes in Tal1−/− ES cells.

Figure Figure 4..

Several genes expressed in mast cells, erythrocytes, and megakaryocytes were not detected in PU.1-on TUNE-1 embryonic stem (ES) cells. After ES cells were cultured on OP9 cells for 5 days, mRNA was purified. The expression of several genes was analyzed by reverse transcription-polymerase chain reaction (RT-PCR). J1: cultured cells from wild-type J1 ES cell line; Tc+: TUNE ES cells cultured in the presence of tetracy-cline (Tc); Tc−: TUNE ES cells cultured with removal of Tc on day 2; OP9: stromal cells; Posi: positive controls.

Gene Expression of the B-Cell Lineage Was Not Detected in the Hematopoiesis from TUNE ES Cells

PU.1 is considered essential for B lymphopoiesis [25]. To assess whether exogenous PU.1 expression was able to rescue B lymphopoiesis from TUNE ES cells, the ES cells were cultured on OP9 stromal cells for 8 days, and then re-cultured on ST2 cells for an additional 10 days. Flt3-ligand (20 ng/ml) and IL-7 (20 U/ml) were added to the cultures during days 5–8, and days 8–18, respectively (Fig. 1E) [26].

Clusters of small cells were observed in parental J1 ES cell cultures on day 18, but no clusters appeared in cultures of TUNE ES cells with or without Tc from day 2 (Fig. 5A). B-lineage–expressed genes were monitored by RT-PCR. Pax5 and Igb were not expressed on cells from TUNE ES cells even in the absence of Tc, although these genes were detected in the cultured cells from wild-type J1 ES cells (Fig. 5B). These results indicate that B-lineage cells are not induced from Tal1−/− ES cells with enforced expression of PU.1 under these culture conditions.

Figure Figure 5..

B-lymphocyte–related genes were not expressed in PU.1-on TUNE-1 embryonic stem (ES) cells. B-lineage cells were induced in the schedule as described in Figure 1E (B cell culture). (A): The appearance of each culture on day 18. (B): mRNA was prepared on day 18 of cultures, and the expressions of each gene were detected by reverse transcription-polymerase chain reaction (RT-PCR). J1: cultured cells from wild-type J1 ES cell line; Tc+: cells cultured in the presence of tetracycline (Tc); Tc−: cells cultured with removal of Tc on day 2; ST2: stromal cells; Posi: positive controls.

Efficiency of Osteoclastogenesis from Tal1−/− ES Cells Depended on the Timing of the Expression of PU.1

Since the Tc-off system permits temporal regulation of PU.1 expression in TUNE ES cells, we examined the relationship between the efficiency of osteoclastogenesis and the timing of PU.1 expression. After switching the culture conditions to induce differentiation on OP9 cells, Tc was withdrawn from cultures of TUNE-1 and TUNE-2 ES cells on successive days. Withdrawal of Tc on days 0–4 allowed osteoclastogenesis to take place more efficiently than withdrawal after 6 days (Fig. 6A). No osteoclasts were induced in cultures of either ES clone subjected to the continuous exposure to Tc. Some variability in efficiency was observed in each experiment and clone, but the trend that day 2 to day 4 was the appropriate timing to express PU.1 for induction of TRAP+ cells from Tal-1–deficient ES cells was reproducible in all experiments and clones. These results indicate that the timing of expression of PU.1 is an important parameter in osteoclast induction, and day 2 to 4 is the appropriate timing. Since endogenous expression of PU.1 in cultures of undifferentiated wild-type J1 ES cells was first detected on day 5 rather than day 3 (data not shown), the timing of PU.1 expression for the rescue of osteoclastogenesis in Tal1−/− ES cells preceded the normal pattern by 1–2 days.

Figure Figure 6..

Different efficiencies of osteoclastogenesis from Tal1−/− ES cells depended on timing of expression of PU.1. (A): TUNE-1 and TUNE-2 ES cells were cultured in the schedule as described in Figure 1E, osteoclast culture. Shaded bars represent periods of addition of Tc. Each column represents the mean number ± SD of TRAP-positive cells in triplicate cultures. (B): The amount of gene expression of Sfpi1 (PU.1) in TUNE-2 cells was not lower than that in cells derived from J1 ES cells. On day 5 of osteoclast culture, mRNA was purified, and semi-quantitative RT-PCR using sequential doses of cDNA (1, 1/2, 1/4) was performed. To estimate the amount of mRNA derived from ES cells, Kdr (Flk-1) RT-PCR products from serially doses of cDNA were indicated. Tc−: cells cultured without Tc; Tc +: cells cultured with Tc.

Efficiency of osteoclastogenesis from TUNE cells was significantly lower than that from wild-type J1 ES cells. To determine whether this resulted from the low expression of PU.1, semi-quantitative RT-PCR was performed on day 5 of osteoclast culture. Expression of the Kdr (Flk-1) gene was used as the control of amount of ES cell–derived mRNA because frequency of Flk-1–expressing cells was not different among J1, PU.1-on, and PU-1-off TUNE cells, and Flk-1 was not expressed in OP9 stromal cells. Among three TUNE clones, TUNE-2 cells were the most efficiently differentiated to osteoclasts, but the efficiency was less than one-hundredth of that from wild-type ES cells. Interestingly, the expression of PU.1 was quantitatively not different from J1 cells (Fig. 6B). These results suggested that an adequate amount of expression of PU.1 was not enough to completely rescue osteoclastogenesis from Tal1−/− ES cells.

Flk-1+ Cells Efficiently Differentiated to Osteoclasts by PU.1 Expression

Based on the results presented above, we hypothesized that hematogenic cells—for example, hemangioblasts and mesodermal cells immediately before their commitment to the hematopoietic cell lineage—might differentiate into osteoclasts with enforced expression of PU.1. Hematopoietic cells appear in vivo on embryonic day 7.5, whereas Kit+ blood cells are first observed on day 4 of ES cell culture [27]. At this time, Flk-1+ cells containing hematopoietic precursors are present in ES cell cultures [28]. To assess the candidate cell populations with potential to differentiate into osteoclasts upon PU.1-expression, we compared the efficiency of osteoclastogenesis in Flk-1+ and Flk-1 fractions.

J1 and TUNE ES cells were differentiated on OP9 cells. On day 3, Tc was withdrawn to allow for expression of PU.1. On day 4, the cells were harvested. No significant difference in the percentage of Flk-1+ cells on day 4 was observed in cells derived from wild-type versus TUNE ES cells, either in the PU.1-off or -on condition (Fig. 7A). We separated the harvested cells into the column-bound (positive) and -passed (negative) fractions by magnetic cell sorting using anti–Flk-1 antibody. The negative fraction contained 0% to 0.05% Flk-1+ cells (Fig. 7A). The fractionated cell populations were induced to form osteoclasts on ST2 cells supplemented with 1α,25(OH)2D3 and Dex for 6 days. The number of TRAP+ cells from the positive fraction of J1 ES cells was increased >200-fold over the number from the negative fraction. In TUNE ES cells expressing PU.1 from day 3, TRAP+ cells developed from the positive fraction and not the negative fraction. Neither fraction of PU.1-off TUNE-1 ES cells generated TRAP+ cells (Fig. 7B). These results suggest that Flk-1+ cells on day 4 of osteoclast culture have the potential to differentiate to osteoclasts in response to PU.1 expression in the absence of Tal-1.

Figure Figure 7..

Flk-1+ cells included osteoclast precursors of PU.1-on TUNE-1 embryonic stem (ES) cells. After ES cells were cultivated on the OP9 layer for 4 days, the cells were harvested and subjected to magnetic sorting using Flk-1 antibody. (A): The proportion of Flk-1+ cells was analyzed by a flow cytometer. Negative: the column-passed fractions by the magnetic cell sorting using anti–Flk-1 antibody. Positive: the column-bound fraction. (B): Osteoclasts were induced from the sorted cells cultured on ST2 cells in the presence of 1α,25(OH)2D3 and dexamethasone (Dex) for 6 days. Each column represents the mean numbers ± SD of artrate-resistant acid phosphatase (TRAP)–positive cells in triplicate cultures. Tc: cells cultured in the presence of tetracycline (Tc).


Tal1 is essential for the development of all hematopoietic lineages and HSCs [1, 2]. The events downstream in the regulatory hierarchy from Tal-1 to specific lineages are largely unknown. Presumably, Tal-1 participates in the transcriptional control of critical downstream factors, perhaps including PU.1. In this study, we demonstrated that enforced PU.1 expression rescues osteoclastogenesis from Tal1−/− ES cells, even though its efficiency was lower than that from J1 ES cells. These results suggest that PU.1 may serve as a critical factor for the osteoclast lineage. In addition, they raise the possibility that one role of Tal-1 may be to activate expression of PU.1 in a regulatory hierarchy. In these experiments, the expression of the endogenous Sfpi1 locus was not activated by exogenous PU.1 (Fig. 4). These observations suggest that PU.1 does not participate in a positive autoregulatory loop; they also indicate that the chromatin structure surrounding the endogenous Sfpi1 gene may be inaccessible before Tal-1 is expressed.

Macrophage-like phagocytes were also induced from PU.1-on TUNE cells. A small number of blood cell–like round cells appeared when PU.1-on TUNE cells were cultured in IL-3, GM-CSF, or G-CSF. These cells were cytospun and observed under microscopy. They looked like monocytes but not granulocytes (data not shown). We prefer the possibility that not only PU.1 but also other transcription factor(s) induced by Tal-1 were needed for granulopoiesis, even though we cannot rule out the possibility that it might be a result of the low sensitivity of granulocyte detection.

In our B-lymphoid culture system, the expression of B-cell–related genes was not detected from PU.1-expressing TUNE ES cells. As shown in Figure 5A, few, if any, hematopoietic-like cells were observed in PU.1-on TUNE ES cell cultures. Since GFP+ cells, indicative of ES cell derivatives (data not shown), and PU.1 transcripts were detected by RT-PCR (Fig. 5B), ES-derived cells were present, though unable to undergo B lymphopoiesis. Pax5-deficient mice have pro-B cells, and the Igb gene is detected [29]. Neither gene was expressed in PU.1-on TUNE ES cell cultures. Thus, the number of B cells generated might have been very small because of the failure of proliferation, or because B-lineage differentiation might need PU.1 and other transcription factor(s). We have not excluded the possibility that the level or timing of PU.1 expression may not have been suitable for B lymphopoiesis under our conditions (Tc withdrawal on day 2).

The efficiency of osteoclastogenesis of PU.1-on TUNE ES cells was significantly lower than that of wild-type ES cells. On culture day 5, the frequency of osteoclast precursors from wild-type ES cells was 1/2,485, while that from PU.1-on TUNE-1 ES cells was 1/11,389 to 1/39,063. We had expected the low dosage of endogenous PU.1 compared with J1 ES cells might be responsible for the incomplete rescue of osteoclastogenesis. However, a comparable amount of PU.1 was detected in TUNE cells (Fig. 6B). This suggests that differentiation of osteoclasts was induced by PU.1, but orchestration of Tal-1 or other genes induced by Tal-1 was needed for efficient osteoclastogenesis.

We cannot rule out the possibility that the timing of the expression of PU.1 may be inappropriate. We showed in Figure 6 that the timing was critical for osteoclastogenesis from PU.1-on TUNE-1 and TUNE-2 ES cells. By regulating the time of expression of PU.1, we found that days 2–4 were most appropriate for rescue of osteoclastogenesis from Tal1-null ES cells. The efficiency of osteoclastogenesis by removal of Tc from day 0 was lower than that from day 2–4. This might show that exogenous PU.1 expression from initiation of cultures affected other than hematopoietic cell lineages. Moreover, according to the decline of osteoclastogenesis by removal of Tc after day 4, these cells might not be maintained in these cultures. Since mesodermal derivatives or hemangioblasts were thought to appear on day 4 of our culture system, they might be candidates of cells competent to differentiate into osteoclasts. These cells might exist in the richest on day 4 of all culture days. Since osteoclasts are of hematopoietic origin, we infer that PU.1 expression in Tal1-null ES cells bypasses a developmental block in a mesodermal derivative or hemangioblast. This interpretation is consistent with our finding that Flk-1+, rather than Flk-1, cells are the source of osteoclasts generated upon PU.1 expression. The development of Flk-1+ cells in the absence of Tal-1 has been reported previously [30].

Tal-1 is indispensable for normal osteoclastogenesis [22]. The results presented here raise the possibility that Tal-1 may act during a brief time interval to establish the hematopoietic program. Once it has activated critical downstream transcription factors, its role in the generation of selected lineages might be dispensable. This is consistent with the observation that Tal-1 is largely dispensable for maintenance of the adult hematopoietic system, once it has been established, except for differentiation of erythroid precursors and megakaryocytes [31]. Ectopic expression of transcription factors, such as illustrated here, may provide a means to define critical factors acting downstream of Tal1 in the hematopoietic regulatory hierarchy.


We acknowledge Dr. Tomohiro Kurosaki for his warm encouragement. We also thank Drs. Shin-Ichi Nishikawa (Riken)foranti-Fmsandanti–Flk-1antibodies;HitoshiNiwa (Riken) and Richard Maki (Burnham Institute) for plasmids; Takao Taki and Masayuki Takahashi (Otsuka Pharmaceutical) for M-CSF; Tetsuo Sudo (Toray Industries) for GM-CSF and IL-3; Takumi Era (Riken) and Toru Nakano (Osaka University) for critical suggestions; Yasuhiko Nagasaka (Beckman Coulter K.K. Tokyo, Japan) for technical assistance of cell sorting by EPICS Elite; and Ms. Toshie Shinohara for her technical support. This work was supported by grants from Grantin-Aid for Scientific Research from the Ministry of Education, Culture, Sports, Science and Technology; from Research on Demential and Fracture, Health and Labour Sciences Research Grants, the Japanese Government (S.I.H, H.Y); and from the Molecular Medical Science Institute, Otsuka Pharmaceutical Co., Ltd. M.T. is a Research Fellow of the Japan Society for the Promotion of Science.