Hematopoietic Stem Cells

Abstract

Hematopoietic stem cells (HSCs) have the capacity to self-renew and the potential to differentiate into all of the mature blood cell types. The ability to prospectively identify and isolate HSCs has been the subject of extensive investigation since the first transplantation studies implying their existence almost 50 years ago. Despite significant advances in enrichment protocols, the continuous in vitro propagation of human HSCs has not yet been achieved. This chapter describes current procedures used to phenotypically and functionally characterize candidate human HSCs and initial efforts to derive permanent human HSC lines.

Introduction

Hematopoietic stem cells (HSCs) are multipotent precursors that have self-renewal capacity and the ability to regenerate all of the different cell types that comprise the blood-forming system (Bonnet, 2002; McCulloch and Till, 2005). Transplantation of HSCs forms the basis of consolidation therapy in cancer treatments and is used to cure or ameliorate a number of hematologic and genetic disorders (Shizuru et al., 2005; Steward and Jarisch, 2005). With certain caveats (McCormack and Rabbitts, 2004), HSCs are also an attractive target cell population for gene therapies because they are readily accessible for ex vivo genetic modification and allow for the possibility of sustained transgene expression in circulating peripheral blood cells throughout the lifetime of an individual (Hawley, 2001; Moayeri et al., 2005).

Historically, mouse HSCs were identified retrospectively by utilizing clonal in vivo assays wherein labeled cells (e.g., genetically-tagged with reporter genes) were assessed for potential to functionally reconstitute hematopoiesis following injection into conditioned hosts, with self-renewal capacity demonstrated by serial transfer into secondary recipients (Abramson et al., 1977; Capel et al., 1990; Jordan and Lemischka, 1990; Keller et al., 1985). Limiting dilution analysis of total bone marrow preparations allowed quantitative estimation of HSC frequencies ranging from 1 in 10,000 to 1 in 100,000 cells (Harrison et al., 1993; Harrison, 1980; Szilvassy et al., 1990). A major advance in the field of HSC biology was the prospective isolation of enriched populations of mouse HSCs based on cell surface phenotype (Spangrude et al., 1988). With the exception of clinical gene marking trials (Stewart et al., 1999), analogous HSC transplantation experiments cannot be performed in humans. For this reason, xenogeneic transplant models have been developed as surrogate assays to evaluate human hematopoietic precursors for in vivo repopulating potential. These assays have helped to elucidate the composition of the human HSC compartment (Bhatia et al., 1998; Gallacher et al., 2000; Glimm et al., 2001; Guenechea et al., 2001; Mazurier et al., 2003; Wang et al., 2003; Zanjani et al., 1998) and have provided paradigms for translation to clinical applications (Baum et al., 1992; Civin et al., 1996b; Lang et al., 2004; Shizuru et al., 2005; Shmelkov et al., 2005; Shpall et al., 1994; Yin et al., 1997).

In this chapter, we discuss the phenotypic and functional characteristics of mouse and human HSCs, and describe protocols for the isolation and assay of candidate human HSCs. A procedure to derive factor-dependent human hematopoietic progenitor cell lines is also provided.

Identification and Enrichment of HSCs

Cell Surface Markers

All HSC activity in adult mouse bone marrow is contained in a population of cells characterized by expression of the c-Kit tyrosine kinase receptor (the receptor for stem cell factor, SCF), stem cell antigen-1 (Sca-1, Ly-6A/E), low levels of the Thy-1.1 cell surface antigen (Thy-1.1^lo), and no or low levels of expression of many cell surface antigens found on differentiated cells belonging to various lineages (referred to as lineage-negative or Lin⁻) (Shizuru et al., 2005). Mouse HSCs variably express the sialomucin CD34, depending on developmental stage and cell cycle status (Ito et al., 2000; Matsuoka et al., 2001; Osawa et al., 1996; Sato et al., 1999). Recent studies have identified a number of additional cell surface antigens that mark mouse HSCs including: the Tie family of receptor tyrosine kinases (Arai et al., 2004; Iwama et al., 1993); endoglin, an ancillary transforming growth factor-β receptor (Chen et al., 2002); endomucin, a CD34-like sialomucin (Matsubara et al., 2005); CD150, the founding member of the SLAM family of receptors (Kiel et al., 2005; Yilmaz et al., 2006); CD201, the endothelial protein C receptor (Balazs et al., 2006); and prion protein (Zhang et al., 2006b). The receptor for thrombopoietin (TPO), c-Mpl, is also expressed on ~70% of mouse c-Kit⁺Lin⁻Sca-1⁺ HSCs (Solar et al., 1998).

In humans, clinical protocols involving enrichment for HSCs generally utilize cells expressing CD34 (Civin et al., 1996b; Shizuru et al., 2005; Shpall et al., 1994), which is expressed on ~0.2–3% of the nucleated cells in cord blood, bone marrow and mobilized peripheral blood (Civin et al., 1984; Krause et al., 1996; Sutherland et al., 1996). Experimentally, further isolation and characterization of Lin⁻ CD34⁺ subpopulations have defined more primitive precursors with hematopoietic repopulating activity that express combinations of the CD59 surface antigen related to Sca-1, the vascular endothelial growth factor receptor-2 (VEGFR2 or KDR), and low levels of c-Kit (CD117), Thy-1 (CD90), and the CD38 surface antigen (Baum et al., 1992; Civin et al., 1996a; Gunji et al., 1993; Hill et al., 1996; Kawashima et al., 1996; Larochelle et al., 1996; Ziegler et al., 1999). As in the mouse, the TIE family of receptor tyrosine kinases and the TPO receptor c-Mpl also appear to further enrich for human HSCs, being expressed on ~80% and ~70% of CD34⁺CD38⁻ cells, respectively (Hashiyama et al., 1996; Ninos et al., 2006; Solar et al., 1998).

It has recently become appreciated that the CD133 cell surface antigen is another important human HSC marker (de Wynter et al., 1998; Gallacher et al., 2000; Hess et al., 2006; Lang et al., 2004; Shmelkov et al., 2005; Yin et al., 1997). CD133, the human homolog of mouse Prominin-1 (Shmelkov et al., 2005), was first identified as a selective human HSC surface molecule using a monoclonal antibody recognizing a particular glycosylated form of Prominin-1 designated as AC133 (Yin et al., 1997). Selection for CD133⁺ hematopoietic precursors yields >90% CD34⁺ cells and contains all of the human hematopoietic repopulating activity. Notably, the extremely rare CD34⁻ candidate HSCs that had previously been identified (Bhatia et al., 1998; Gao et al., 2001; Wang et al., 2003; Zanjani et al., 1998) reside within the CD133 fraction (Gallacher et al., 2000).

Enriched populations of human HSCs are routinely obtained by positive selection for CD34/CD133 and/or by depletion of lineage-committed cells using monoclonal antibodies recognizing differentiation markers (such as CD2, CD3, CD14, CD16, CD19, CD24, CD41, CD56, CD66b and CD235a) in the context of immunomagnetic or fluorescence-activated cell sorting methodologies. In this regard, it is important to bear in mind that physical manipulation of HSCs during the enrichment procedure may not be without effect on cell physiology (Kimura et al., 2004). For example, it is conceivable that binding of antibodies to CD34/CD133 may trigger intracellular signaling pathways that could modulate HSC function. Interestingly, a recent study suggests that the majority of cells within the CD34⁺CD38⁻Lin⁻ HSC compartment express the myeloid-associated lineage markers CD13, CD33 and CD123 (the low affinity binding subunit of the IL-3 receptor) (Taussig et al., 2005), indicating that some caution is warranted when selecting a cocktail of monoclonal antibodies for lineage marker-depletion enrichment of human HSCs.

Fluorescent Dye Staining

Hoechst 33342 and Rhodamine 123

Other strategies that have been utilized to identify and enrich for HSCs are based on the staining patterns of fluorescent dyes (Bertoncello et al., 1985; Goodell et al., 1996; Jones et al., 1995; Leemhuis et al., 1996; Storms et al., 1999; Visser et al., 1981; Wolf et al., 1993). Rhodamine 123 (which preferentially accumulates in active mitochondria) and Hoechst 33342 (a bis-benzimidazole that binds to adenine-thymine rich regions of the minor groove of DNA) are two fluorescent vital dyes that have been routinely used to characterize hematopoietic precursor populations (Bertoncello et al., 1985; Leemhuis et al., 1996; McAlister et al., 1990; Visser et al., 1981; Wolf et al., 1993). Rhodamine 123 staining of mouse bone marrow cells demonstrated that HSCs with long-term repopulating potential stained dimly whereas more brightly staining hematopoietic precursors could only provide short-term repopulation (Bertoncello et al., 1988; Bertoncello et al., 1991; Spangrude and Johnson, 1990; Zijlmans et al., 1995). Moreover, subpopulations of mouse bone marrow cells that stained most weakly with both dyes were shown to be highly enriched for long-term repopulating HSCs (Bertoncello and Williams, 2004; Leemhuis et al., 1996; Wolf et al., 1993). Decreased staining with these dyes generally reflects a metabolically and mitotically inactive state (Arndt-Jovin and Jovin, 1977; Johnson et al., 1980; Spangrude and Johnson, 1990). However, it is now appreciated that decreased staining of HSCs with rhodamine 123 and Hoechst 33342 is also due to efflux mediated by at least two members of the ATP-binding cassette (ABC) family of transporters, ABCB1 (also referred to as MDR1 or P-glycoprotein) and ABCG2 (also referred to as BCRP, MXR or ABCP) (Chaudhary and Roninson, 1991; Juliano and Ling, 1976; Scharenberg et al., 2002; Zhou et al., 2001; Zhou et al., 2002).

Side Population Assay

A novel method that simultaneously monitors low fluorescence intensity of Hoechst 33342 staining at ~450 nm and at >675 nm following ultraviolet excitation identifies a rare (<0.1%) subpopulation of mouse bone marrow cells, referred to as ‘side population’ (SP) cells, which contains the vast majority of long-term hematopoietic repopulating activity (Goodell et al., 1996). The ABC transporter Bcrp1 (the mouse ortholog of human ABCG2) expressed in mouse bone marrow cells is the major determinant of the mouse SP profile (Zhou et al., 2001; Zhou et al., 2002). Subsequent multiparameter flow cytometric analysis of mouse bone marrow SP cells showed that approximately one-third exhibited the c-Kit⁺Thy-1.1^loLin⁻Sca-1⁺ phenotype while approximately one-half expressed CD34 (Pearce et al., 2004). In another study, mouse bone marrow cells with the strongest dye efflux activity, which exhibited the highest hematopoietic repopulating activity, were shown to have a c-Kit⁺Lin⁻Sca-1⁺CD34⁻ phenotype (Matsuzaki et al., 2004).

The SP assay has also been applied to human hematopoietic tissues (Eaker et al., 2004; Goodell et al., 1997; Naylor et al., 2005; Preffer et al., 2002; Scharenberg et al., 2002; Storms et al., 2000; Uchida et al., 2001). Unlike mouse bone marrow SP cells, human hematopoietic SP cells constitute a much more phenotypically and functionally heterogeneous precursor population (Naylor et al., 2005; Preffer et al., 2002; Storms et al., 2000; Uchida et al., 2001). CD34⁻ SP cells have been identified in several studies, but to date repopulating ability of human hematopoietic SP cells has only been demonstrated for CD34⁺ subpopulations (Eaker et al., 2004; Scharenberg et al., 2002; Uchida et al., 2001).

Fluorescent Substrates for Cytosolic Aldehyde Dehydrogenase Activity

Cytosolic aldehyde dehydrogenase (ALDH), an enzyme responsible for oxidizing a variety of intracellular aldehydes, is expressed at high levels in HSCs, conferring resistance to the alkylating agents cyclophosphamide and 4-hydroxyperoxycyclophosphamide (Gordon et al., 1985; Kastan et al., 1990; Sahovic et al., 1988). Fluorescent substrates for ALDH have been developed and shown to be useful for isolating mouse and human HSCs (Fallon et al., 2003; Hess et al., 2006; Hess et al., 2004; Jones et al., 1996; Jones et al., 1995; Storms et al., 1999). In proof-of-principle studies (Jones et al., 1996; Jones et al., 1995), dansyl-aminoacetaldehyde (DAAA) was used as an ALDH substrate. DAAA can diffuse freely across the cell membrane because it is uncharged. Cells expressing ALDH oxidize DAAA to dansyl-glycine, which is retained intracellularly by virtue of a charged carboxylate group at physiologic pH, and ALDH⁺ cells are identified by dansyl fluorescence upon excitation with ultraviolet light. More recently, a newer fluorescent substrate for ALDH—termed BODIPY^™-aminoacetaldehyde (BAAA)—was synthesized, which uses a nontoxic visible light-excitable fluorophore BODIPY^™ (Storms et al., 1999). Similar to DAAA, BAAA is uncharged and diffuses freely across the cell membrane, becoming converted to BODIPY^™-aminoacetate (BAA), which is retained intracellularly because of its net negative charge in the presence of an inhibitor of the ABC transporter ABCB1 (Storms et al., 1999).

Mouse hematopoietic precursors enriched for high expression of ALDH by staining with BAAA or DAAA may represent a novel class of HSCs, which express undetectable or low levels of the c-Kit, Thy-1, Sca-1 and CD34 HSC markers, but which produce long-term albeit delayed multilineage engraftment (Armstrong et al., 2004; Jones et al., 1996). Flow cytometric analysis of human cord blood cells stained with BAAA identified a population of cells (at a frequency of ~1%) with bright fluorescence intensity (ALDH^br) and low orthogonal light “side” scattering (SSC^lo) comprising ~74% CD34⁺ cells and ~46% CD34⁺CD38^lo/− cells, which was largely depleted of cells with mature T cell, natural killer cell, myeloid, erythroid and platelet lineage markers (Storms et al., 1999). The SSC^loALDH^br population still contained a small number of B cells however (~12%) (Storms et al., 1999). In another study, Lin⁻ depletion combined with selection for ALDH^br cells by BAAA staining demonstrated enrichment for hematopoietic precursors coexpressing CD133 and CD34 (~73%) and all of the hematopoietic repopulating activity in human cord blood preparations (Hess et al., 2004). A recent follow-up study by the same group reported that prospective selection of ALDH^brLin⁻ human cord blood cells for CD133 expression yields a population of primitive precursors that are primarily CD34⁺ (~95%) and which contains all long-term hematopoietic repopulating activity (Hess et al., 2006).

A two-step enrichment strategy for human HSCs combining positive selection for CD133⁺ cells and assay for high level ALDH expression (SSC^loALDH^br cells) will be described here.

Isolation of Candidate Human HSCs Protocol

1. Obtain human cord blood, bone marrow or mobilized peripheral blood cells after informed consent in conformity with a human subjects protocol approved by an Institutional Review Board, or purchase from a commercial source (e.g., AllCells, Cambrex, StemCell Technologies). For human cord blood cells, dilute anticoagulated cord blood 1:3 with phosphate buffered saline (PBS) containing 0.6% anticoagulant citrate dextrose solution A (ACD-A, Sigma-Aldrich, Catalog number C3821). Layer 35 ml of diluted cord blood over 12 ml of Ficoll-Paque PLUS (GE Healthcare Life Sciences, Catalog number 17-1440-02). Centrifuge at 375 g for 30 min at room temperature (22°). Collect cells at the interface, dilute with PBS containing 0.6% ACD-A and centrifuge at 375 g for 10 min at 22°. Resuspend cells in erythrocyte lysing solution (0.15 M NH₄Cl, 1.0 mM KHCO₃, 0.1 mM EDTA, pH 7.2–7.4) and incubate for 10 min at 22°. Centrifuge at 375 g for 10 min at 22° and wash once in PBS.

2. Subsequent enrichment for cells expressing CD133 can be performed with the CD133 MicroBead Kit (Miltenyi Biotec, Catalog number 130-050-801) utilizing superparamagnetic beads conjugated to a monoclonal mouse anti-human CD133/1 antibody and a VarioMACS^™ Separator (Miltenyi Biotec, Catalog number 130-090-282). Follow the manufacturer’s recommendations and obtain ~1 × 10⁶ CD133⁺ cells/ml with >95% purity (if necessary, repeat the enrichment using a second MACS^® cell separation column).

3. Prepare aliquots of ~5 × 10⁴ CD133⁺ cells/50 ul in PBS containing 2% fetal bovine serum (FBS) for staining individually with fluorochrome-conjugated anti-CD133/2 (which recognizes a different epitope than CD133/1), anti-CD34 and anti-CD38 monoclonal antibodies, as compensation controls: unstained; anti-CD133/2-phycoerythrin (PE) (Miltenyi Biotec, Catalog number 130-090-853); anti-CD34-peridinin chlorophyll protein (PerCP) (BD Biosciences Pharmingen, Catalog number 340430); and anti-CD38-allophycocyanin (APC) (BD Biosciences Pharmingen, Catalog number 555462). Incubate at 4° for 20 min. Add 2 ml PBS containing 2% FBS. Centrifuge at 375 g for 10 min at 4°. Decant supernatant and drain. Resuspend in 300 μl PBS containing 2% FBS. Keep on ice until analysis.

4. Identification of CD133⁺ cells expressing high levels of ALDH activity can be facilitated by using a commercial kit for BAAA staining (StemCell Technologies, ALDEFLUOR^® kit, Catalog number 01700). Centrifuge 1 × 10⁶ CD133⁺ cells at 375 g for 10 min at 25°. Decant supernatant and drain. Resuspend CD133⁺ cells in 1 ml of proprietary ALDEFLUOR^® assay buffer (containing an inhibitor of ABCB1 transporter efflux activity) with the ALDEFLUOR^® reagent (BAAA-DA; BODIPY^™-aminoacetaldehyde diethyl acetal) according to the manufacturer’s instructions. BAAA-DA is dissolved in dimethylsulfoxide and exposed to hydrochloric acid to convert it to the ALDH substrate BAAA. As BAAA diffuses freely across the cell membrane, all of the viable cells will be fluorescent. However, cells with high ALDH activity metabolize the substrate into BAA (BODIPY^™-aminoacetate) containing a charged carboxylate group and become intensely fluorescent. Including an inhibitor of ABCB1 transporter efflux activity throughout the assay ensures retention of the fluorescent BAA compound within the cell. Cells incubated in the presence of diethylaminobenzaldehyde (DEAB), a potent ALDH inhibitor, provide a control for background BAA fluorescence.

5. Upon completion of the assay (30–60 min), stain aliquots of the cells with combinations of anti-CD133/2-PE, anti-CD34-PerCP and anti-CD38-APC at 2–8° as above. After 20 min, centrifuge all samples at 375 g for 10 min at 4°. Decant supernatant and drain. Resuspend samples in ALDEFLUOR^® assay buffer.

6. Analyze on a flow cytometer equipped with excitation wavelengths of 488 nm and 633 nm. Detect scatter and fluorescence signals with: 488/10 bandpass (BP) filters for SSC and forward scatter signals, 530/30 BP for BAA fluorescence, 576/26 BP for anti-CD133-PE, 675/20 BP for anti-CD34-PerCP, and 660/20 BP for anti-CD38-APC signals.

7. Fluorescence activated cell sorting of CD133⁺ cells expressing the highest levels of ALDH activity enriches for candidate human HSCs with a predominantly SSC^loALDH^brCD133⁺CD34⁺CD38^lo phenotype (Fig. 1). Note: Since dead and dying cells without intact cellular membranes cannot retain the fluorescent BAA derivative, only viable cells are identified by this method.

FIG. 1.

Flow cytometric characterization of human CD133⁺ cord blood cells expressing high levels of ALDH activity. (A) Negative control: Human cord blood cells highly enriched for CD133 expression (>95% CD133⁺) exhibiting low side scatter (SSC^lo) stained with BAAA (using the ALDEFLUOR^® reagent) in the presence of DEAB, a potent ALDH inhibitor, show background levels of BAA fluorescence (ALDH^dim). Events to the left of the ALDH^dim gate represent dead cells with no BAA fluorescence. (B) Human cord blood cells highly enriched for CD133 expression (>95% CD133⁺) exhibiting low side scatter (SSC^lo) stained with BAAA (using the ALDEFLUOR^® reagent) in the absence of DEAB showing that almost all of the cells expressed high levels of ALDH activity (ALDH^br). (C) The vast majority of ALDH^brCD133⁺ cells coexpress the CD34 HSC surface antigen. (D–F) Flow cytometric analysis indicates that cells within the more primitive CD133⁺CD34⁺CD38^lo subpopulation express higher levels of ALDH activity than cells within the CD133⁺CD34⁺CD38^hi subpopulation. (D) Gating strategy for CD133⁺CD34⁺CD38^hi and CD133⁺CD34⁺CD38^lo subpopulations. (E) CD133⁺CD34⁺CD38^hi cells are enriched for cells with the lowest levels of BAA fluorescence within the ALDH^br gate. (F) CD133⁺CD34⁺CD38^lo cells are enriched for cells with the highest levels of BAA fluorescence within the ALDH^br gate. Flow cytometry data was acquired on a FACSAria instrument (BD Biosciences) and analyzed with WinList 3D v6.0 pre-release software (Verity Software House).

Functional Characterization of Candidate HSCs

Surrogate In Vivo Assays

Heterogeneity of the human HSC compartment and continued questions regarding cell surface phenotype necessitate the use of in vivo assays of HSC function (Baum et al., 1992; Bhatia et al., 1998; Dao et al., 2003; Dorrell et al., 2000; Gallacher et al., 2000; Glimm et al., 2001; Guenechea et al., 2001; Mazurier et al., 2003; Sieburg et al., 2006; Wang et al., 2003; Zanjani et al., 1998).

Several xenogeneic transplant models have been developed as surrogate assays of human hematopoietic repopulating cells. The preimmune fetal sheep transplant assay has emerged as a useful large animal model (Civin et al., 1996a; Zanjani et al., 1996). However, the majority of functional assays of human HSC activity involve transplantation into immunodeficient mice with varying degrees of residual natural immunity (Bhatia et al., 1997b; Bock et al., 1995; Cashman et al., 1997; Cheng et al., 1998; Gimeno et al., 2004; Glimm et al., 2001; Goldman et al., 1998; Guenechea et al., 2001; Hiramatsu et al., 2003; Hogan et al., 1997; Ishikawa et al., 2002; Ito et al., 2002; Kamel-Reid and Dick, 1988; Kollet et al., 2000; Kyoizumi et al., 1992; Lapidot et al., 1992; Larochelle et al., 1996; Lowry et al., 1996; Mazurier et al., 1999; McCune et al., 1991; Meyerrose et al., 2003; Nolta et al., 1994; Pflumio et al., 1996; Shultz et al., 2005; Traggiai et al., 2004; Vormoor et al., 1994; Wang et al., 1997). The most widely used of these small animal models is the NOD.CB17-Prkdc^scid mouse – nonobese diabetic (NOD) mice crossed with severe combined immunodeficient (SCID) mice (Bhatia et al., 1997b; Cashman et al., 1997; Hogan et al., 1997; Larochelle et al., 1996; Lowry et al., 1996; Pflumio et al., 1996; Shultz et al., 1995; Wang et al., 1997). NOD/SCID mice support human cell engraftment due to defective rearrangement of T cell receptor and immunoglobulin (Ig) genes, resulting in defects of functional T and B cells; they also have low levels of natural killer cell cytotoxic activity, functionally immature macrophages and an absence of hemolytic complement. Candidate HSCs collectively termed SCID-repopulating cells (SRCs) are scored positive for engraftment if ~1% CD45⁺ human cells or >0.1% human DNA can be detected in the bone marrow of NOD/SCID recipients at or greater than 6 weeks post-transplantation. Under most conditions, the NOD/SCID xenograft assay does not require administration of exogenous human cytokines; however, a sublethal conditioning regimen of 250–400 cGy irradiation is necessary, and cytokine administration or co-administration of accessory cells facilitates engraftment at limiting doses (Bonnet et al., 1999). Under these conditions, the frequency of SRC in human cord blood cells was determined to be 1 in 9.3 × 10⁵ mononuclear cells (Wang et al., 1997) and 1 in 617 CD34⁺CD38⁻Lin⁻ cells (Bhatia et al., 1997b). While both lymphoid and myeloid cell populations are found, a shortcoming of the NOD/SCID xenograft assay is the general lack of T cell development, and differentiation of human hematopoietic precursors is limited mainly to immature cells belonging to the B-cell and, to a lesser degree, myeloid lineages. Other disadvantages of the NOD/SCID mouse model include its high sensitivity to irradiation and relatively short life span (~80% of female and ~50% of male NOD/SCID mice develop lethal thymic lymphomas by 20 weeks of age).

Attempts to obtain an improved host for human HSC transplantion led to the development of a strain of immunodeficient mice in which the residual low natural killer activity present in the NOD/SCID mouse was eliminated by backcrossing the β2 microglobulin null allele onto the NOD/SCID background (NOD/SCID/B2m^−/−) (Kollet et al., 2000). NOD/SCID/B2m^−/− mice support a more than 11-fold higher level of SRC frequency than NOD/SCID mice, with transplantation of ~8 × 10⁴ human cord blood mononuclear cells resulting in mutlilineage differentiation in the mouse bone marrow (Kollet et al., 2000). The enhanced SRC frequency in NOD/SCID/B2m^−/− mice is due to short-term repopulation by myeloid-restricted CD34⁺CD38⁺ cells and a predominantly CD34⁺CD38⁻ population that has broader lymphomyeloid differentiation potential but which does not efficiently engraft NOD/SCID mice (Glimm et al., 2001). A limitation of NOD/SCID/B2m^−/− mice is a relatively short life span due to earlier onset and increased incidence of thymic lymphomas (the mean life span of NOD/SCID/B2m^−/− mice is ~11 weeks shorter than NOD/SCID mice) (Christianson et al., 1997). Recently, new NOD/SCID models for human HSC engraftment have been reported that lack a functional X-linked common cytokine receptor γ-chain gene (NOD/SCID/γ_c⁻) (Ito et al., 2002; Shultz et al., 2005; Yahata et al., 2002). NOD/SCID/γ_c⁻ mice support ~6-fold higher percentages of human hematopoietic cells in the host bone marrow than NOD/SCID mice, with precursors developing into mature human CD3⁺CD4⁺ and CD3⁺CD8⁺ T cells, Ig⁺ B cells, natural killer cells, myeloid cells and plasmacytoid dendritic cells. Notably, NOD/SCID/γ_c⁻ mice survive beyond 16 months of age and even after sublethal irradiation resist lymphoma development.

Other immunodeficient mouse models have been created by crossing mice with a deficient recombinase activating gene 2 (Rag2) with mice harboring the γ_c cytokine receptor gene deletion (Gimeno et al., 2004; Goldman et al., 1998; Mazurier et al., 1999; Traggiai et al., 2004; Weijer et al., 2002). Rag2^−/−γ_c⁻ mice are characterized by absence of all T cell, B cell and natural killer cell function and show no spontaneous lymphoma development. However, efficient human multilineage hematopoietic engraftment in Rag2^−/−γ_c⁻ mice with a mixed H-2 major histocompatibility locus background requires exogenous human cytokines (Mazurier et al., 1999).

As noted earlier, the TPO receptor c-Mpl is a selective marker of mouse and human HSCs (Hashiyama et al., 1996; Ninos et al., 2006; Solar et al., 1998). Consistent with this observation, TPO has been demonstrated to be an important HSC supportive factor (Alexander et al., 1996; Fox et al., 2002; Kaushansky, 2003a; Petzer et al., 1996; Solar et al., 1998) in addition to being the physiologic regulator of megakaryocytopoiesis and thrombopoieis (Kaushansky, 2003b). A recent report suggested that human TPO is a major limiting factor for multilineage outgrowth of human hematopoietic cells in NOD/SCID mice (Verstegen et al., 2003). To assess the effects of human TPO on hematopoietic engraftment of candidate human HSCs in Rag2^−/−γ_c⁻ mice, we generated human TPO-producing Rag2^−/−γ_c⁻ mice by lentiviral vector-mediated transgenesis (Lois et al., 2002; Ma et al., 2003; Pfeifer et al., 2002; Punzon et al., 2004). A self-inactivating (SIN) HIV-1-based lentiviral vector, SINF-EF-hTPO-W, was developed that expresses the human TPO cDNA from an internal human elongation factor 1a (EF1α) promoter (Ramezani et al., 2000; Ramezani et al., 2003; Ramezani and Hawley, 2002a; Ramezani and Hawley, 2003). Concentrated vesicular stomatitis virus G glycoprotein (VSV-G)-pseudotyped lentiviral vector particles (10⁸ transducing units/ml) were microinjected into the perivitelline space of single-cell H-2^b Rag2^−/−γ_c⁻ embryos and implanted into pseudopregnant recipient H-2^b Rag2^−/−γ_c⁻ mice (Lois et al., 2002; Punzon et al., 2004; Ramezani and Hawley, 2002b). Polymerase chain reaction analysis of genomic tail DNA using a forward primer located within the EF1α promoter and a reverse primer located within the human TPO cDNA was used to detect founder animals carrying the integrated transgene (H-2^b Rag2^−/−γ_c⁻-hTPO mice). Serum levels of human TPO in the founder mice ranged between 100–500 pg/ml (R. Behnam, M.B. Chase, S. Soukharev, A.R., and R.G.H., unpublished data). Human CD34⁺ hematopoietic cells were isolated from cord blood as described below and intravenously injected into sublethally irradiated (350 cGy) H-2^b Rag2^−/−γ_c⁻-hTPO and control H-2^b Rag2^−/−γ_c⁻ mice. As a potential preclinical predictor of the rate of platelet recovery after transplantation and thus an indication of the quality of hematopoietic engraftment (Angelopoulou et al., 2004; Bruno et al., 2004; Perez et al., 2001; Yasui et al., 2003), human platelets were evaluated in peripheral blood from weeks 1 to 8 after transplantation. Human platelets were detected in the peripheral blood of H-2^b Rag2^−/−γ_c⁻-hTPO but not control H-2^b Rag2^−/−γ_c⁻ mice by week 3 (1.2 ± 0.8%), reaching 8 ± 2% at week 8 (Fig. 2). Flow cytometric analysis of nucleated peripheral blood cells revealed that all of the H-2^b Rag2^−/−γ_c⁻-hTPO mice (15/15) but none of the control H-2^b Rag2^−/−γ_c⁻ mice (0/6) engrafted with human hematopoietic cells (17 ± 7% CD45⁺ human cells at 6 weeks post-transplantation; Fig. 3A). Slightly higher engraftment levels were obtained in mice that received co-administration of CD34⁻Lin⁺ accessory cells (17% vs 13%). Of the engrafted CD45⁺ human hematopoietic cells, 13 ± 2% were CD19⁺ cells belonging to the B cell lineage and 26 ± 4% were CD33⁺ myeloid cells (Fig. 3B). In contrast to the negative results obtained with adult H-2^b Rag2^−/−γ_c⁻ recipients, transplantation of CD34⁺ human hematopoietic progenitor cells into sublethally irradiated H-2^d Rag2^−/−γ_c⁻ newborns leads to de novo development of T cells, B cells, natural killer cells, myeloid cells and plasmacytoid dendritic cells, formation of structured primary and secondary lymphoid organs, and production of functional immune responses (Gimeno et al., 2004; Traggiai et al., 2004).

FIG. 2.

Human platelet production in peripheral blood of Rag2^−/−γ_c⁻-hTPO mice transplanted with candidate human HSCs. (A) Sublethally irradiated (350 cGy) Rag2^−/−γ_c⁻-hTPO and Rag2^−/−γ_c⁻mice were transplanted with 5 × 10⁵ human CD34⁺ cord blood cells. Human platelets were detected in the peripheral blood of all Rag2^−/−γ_c⁻-hTPO mice but not Rag2^−/−γ_c⁻ mice by staining with an anti-human CD41a monoclonal antibody and gating on low forward and side scatter (platelet population gate). Shown are representative examples. Flow cytometry data was acquired on a FACSCalibur instrument and analyzed with CellQuest software (BD Biosciences). (B) Summary of the analysis of human CD41a⁺ platelets within the platelet population in the peripheral blood of individual Rag2^−/−γ_c⁻-hTPO mice 4 to 8 weeks after transplantation with 5 × 10⁵ human CD34⁺ cells plus (dark bars) or minus (white bars) 1 × 10⁶ CD34⁻Lin⁺ accessory cells.

FIG. 3.

Multilineage human hematopoietic engraftment in the peripheral blood of Rag2^−/−γ_c⁻-hTPO mice transplanted with candidate human HSCs. (A) Sublethally irradiated (350 cGy) mice were transplanted with 5 × 10⁵ human CD34⁺ cord blood cells plus (+) or minus (−) 1 × 10⁶ CD34⁻Lin⁺ accessory cells. Shown is a summary of the percentages of CD45⁺ human cell engraftment in the peripheral blood of transplanted Rag2^−/−γ_c⁻ (open circles) and Rag2^−/−γ_c⁻-hTPO (closed circles) mice 6 weeks after transplantation. Each circle represents data for an individual mouse and the horizontal lines indicate the mean levels of human cells. (B) Flow cytometric analyses showing percentages of human CD45⁺CD19⁺ B cells and CD45⁺CD33⁺ myeloid cells in the peripheral blood of a representative Rag2^−/−γ_c⁻-hTPO mouse at 6 weeks post-transplantation. Flow cytometry data was acquired on a FACSCalibur instrument and analyzed with CellQuest software (BD Biosciences).

Human Hematopoietic Repopulating Cell Assay Protocol

1. Isolate mononuclear cells from human cord blood as described in the previous section. Enrich for cells expressing CD34 with the CD34 MicroBead Kit (Miltenyi Biotec, Catalog number 130-046-703) utilizing superparamagnetic beads conjugated to a monoclonal mouse anti-human CD34 antibody and a VarioMACS^™ Separator. Follow the manufacturer’s recommendations and obtain ~1 × 10⁶ CD34⁺ cells/ml with >95% purity (if necessary, repeat the enrichment using a second MACS^® cell separation column). Retain the CD34⁻Lin⁺ flow-through cells for co-administration as accessory cells.

2. All animal procedures are carried out in accordance with Institutional Animal Care and Use Committee guidelines. H-2^b Rag2^−/−γ_c⁻ ((C57BL/6J × C57BL/10SgSnAi)-[KO]γ_c-[KO]Rag2, Taconic, Catalog number 004111) and NOD/SCID (NOD.CB17-Prkdc^scid, The Jackson Laboratory, Catalog number 001303) immunodeficient mice are housed in sterile microisolator cages on laminar flow racks to minimize the chance of adventitious infections. Two to six hours prior to transplantation, the mice are exposed to a single sublethal dose of total body γ-irradiation from a ¹³⁷Cs source (350 cGy for H-2^b Rag2^−/−γ_c⁻ mice; 250 cGy for NOD/SCID mice). Baytril (Bayer; active ingredient: enrofloxacin) is added to the drinking water (2 ml/250 ml) immediately after irradiation and treatment is continued for 3 weeks as an additional prophylactic measure to prevent possible deaths due to adventitious infections.

3. Prepare aliquots of ~5 × 10⁵ CD34⁺ cells with or without 1 × 10⁶ CD34⁻Lin⁺ cells (as accessory cells) in 200 μl PBS and transplant into sublethally irradiated 8- to 10-week-old immunodeficient mice via intravenous tail vein injection using a 27 gauge needle.

4. Detection of human platelets in mouse peripheral blood: Mouse bleeding (from the retro-orbital venous sinus) is performed after inhalation isoflurane anesthesia and administration to the eye of one drop of a local anesthetic (Tetracaine Ophthalmic Solution, 0.5% solution; Phoenix Pharmaceutical). At weekly intervals post-transplantation, collect peripheral blood from the retro-orbital venous sinus using microhematocrit capillary tubes (Fisher Scientific, Catalog number 22-362-566) and place ~100 μl blood into micro-collection tubes containing potassium ethylenediaminetetraacetic (EDTA; Sarstedt, Catalog number 41.1395.105). Centrifuge at 1,500 g for 5 min at 22° and resuspend the pellet in 500 μl PBS containing 2% FBS. Stain 50 μl aliquots of the cell/platelet suspension for 30 min at 22° with the following monoclonal antibodies: fluorescein isothyocyanate (FITC)-conjugated anti-human CD41a (BD Biosciences Pharmingen, Catalog number 555466) or FITC-conjugated mouse IgG₁ isotype control (BD Biosciences Pharmingen, Catalog number 349041), and anti-mouse CD61-PE (BD Biosciences Pharmingen, Catalog number 553347) or PE-conjugated hamster IgG₁ isotype control (BD Biosciences Pharmingen, Catalog number 553972). Centrifuge at 750 g for 5 min at 22°. Decant supernatant and drain. Wash in 2 ml PBS containing 2% FBS plus 0.1% NaN₃. Centrifuge at 750 g for 5 min at 22°. Decant supernatant and drain. Resuspend in 500 μl PBS containing 2% FBS plus 0.1% NaN₃. Platelets are analyzed on a flow cytometer equipped with excitation wavelengths of 488 nm and 633 nm by gating for low SSC and forward scatter signals (Perez et al., 2001).

5. Detection of human hematopoietic cells in mouse bone marrow: Mice are euthanized under inhalation isoflurane anesthesia by cervical dislocation at or greater than 6 weeks post-transplantation. Single-cell bone marrow suspensions are prepared by flushing the femurs and tibiae with PBS containing 2% FBS using a 21-gauge needle. Erythrocytes are removed by hypotonic lysis in 0.15 M NH₄Cl, 1.0 mM KHCO₃, 0.1 mM EDTA, pH 7.2–7.4 (Eaker et al., 2004; Ramezani et al., 2000). Prepare aliquots of ~1 × 10⁵ cells/100 μl in PBS containing 2% FBS and stain as described in the previous section with the following monoclonal antibodies (all from BD Biosciences Pharmingen): anti-human CD45-FITC (Catalog number 555482), FITC-conjugated mouse IgG₁ isotype control (Catalog number 349041), anti-human CD19-APC (Catalog number 555415), anti-human CD33-APC (Catalog number 551378) and APC-conjugated mouse IgG₁ isotype control (Catalog number 555751). As an additional negative control, stain the bone marrow from an untransplanted mouse. Incubate at 4° for 20 min. Add 1 ml PBS containing 2% FBS plus 0.1% NaN₃. Centrifuge at 375 g for 10 min at 4°. Decant supernatant and drain. Resuspend in 500 μl PBS containing 2% FBS plus 0.1% NaN₃ and analyze by flow cytometry.

Long-term Culture of Candidate HSCs and Progenitors

Overview

Although a variety of culture conditions support some self-renewal of human hematopoietic progenitors, long-term maintenance of HSCs in vitro remains a major challenge to the field (Sauvageau et al., 2004). As illustrated in Fig. 4, a significant decrease in human hematopoietic repopulating activity in NOD/SCID mice was observed following in vitro culture of CD34⁺ cord blood cells for 4 days in serum-free medium supplemented with a combination of SCF, TPO and Flt3 ligand (Larochelle et al., 1996; Petzer et al., 1996). However, modest expansion of SRC (2-to 6-fold net increases) has been reported after short-term culture in serum-free medium in more complex cocktails of hematopoietic growth factors (Bhatia et al., 1997a; Conneally et al., 1997; Gammaitoni et al., 2003). Evidence for a limited degree of in vitro expansion of candidate human HSCs has also been obtained under other culture conditions (Ando et al., 2006; Chute et al., 2005; Madlambayan et al., 2005; Piacibello et al., 1999; Ueda et al., 2000).

FIG. 4.

Loss of human hematopoietic repopulating potential during short-term in vitro culture of CD34⁺ cord blood cells. The potential of human CD34⁺ cord blood cells to engraft in the bone marrow of NOD/SCID mice was compared for cells cultured in vitro for 1 or 4 days in X-VIVO-15 serum-free medium supplemented with 10% BIT 9500 serum substitute, 100 μM β-mercaptoethanol, 100 ng/ml SCF, 20 ng/ml TPO and 100 ng/ml Flt3 ligand. The cells (1.5 × 10⁶) were harvested, mixed with 1 × 10⁶ CD34⁻Lin⁺ accessory cells and transplanted into sublethally irradiated (250 cGy) NOD/SCID mice. Twelve weeks after transplantation, the mice were euthanized and bone marrow cells collected for flow cytometric analysis. Human cells in the mouse bone marrow were detected after staining with anti-human CD45-PE-Cy5 (BD Biosciences Pharmingen, Catalog number 555484) monoclonal antibody. Transplantation of mice with CD34⁺ cord blood cells after 1 day of in vitro culture resulted in ~5% (0.5–50%) human hematopoietic cell engraftment (closed circles). The ability to repopulate NOD/SCID mouse bone marrow was significantly reduced (0–8%, mean: 0.2%) when the CD34⁺ cord blood cells were cultured in vitro for 4 days (open circles).

We have recently explored another approach based on the observation that human embryonic stem cells circumvent cellular senescence by expressing the catalytic subunit of telomerase reverse transcriptase, hTERT, a specialized ribonucleoprotein complex that is responsible for adding telomeric DNA (repetitive TTAGGG sequences) to the ends of chromosomes to prevent shortening during replication (Smogorzewska and de Lange, 2004; Thomson et al., 1998). Candidate human HSCs express relatively high levels of hTERT (Yui et al., 1998), and telomere length analysis of human HSC subpopulations indicates that cells with the longest telomeres have the greatest proliferative potential (Bartolovic et al., 2005; Van Ziffle et al., 2003). Conversely, patients with aplastic anemia have short telomeres and mutations in telomerase have been identified as the cause of hematopoietic failure (Vulliamy et al., 2001; Yamaguchi et al., 2005). Besides progressive telomere shortening, human cells undergo senescence in response to various types of stress (Campisi, 2005). Regardless of the senescence-initiating stimuli, the signaling pathways triggered converge to varying extents on the p53 and retinoblastoma (Rb) tumor suppressors. Therefore, we employed HIV-1-based SIN lentiviral vectors to introduce the hTERT gene and the human papillomavirus type 16 (HPV16) E6 and E7 genes (Okamoto et al., 2002), which accelerate the degradation of p53 and Rb, respectively (Munger et al., 2004), into human CD34⁺ cord blood cells. The transduced CD34⁺ cells were then maintained under serum-free conditions in the presence of SCF, TPO and Flt3 ligand, with or without IL-3 (Akimov et al., 2005). Although this strategy did not result in the immortalization of human HSCs, several SCF-dependent cell lines resembling human myeloerythroid/mast cell progenitors were established in this manner, two of which express low levels of the HSC surface antigen CD133. It is important to point out, however, that the cell lines contain chromosomal aberrations (Table 1). Abnormal karyotypes notwithstanding, the progenitor cell lines were not leukemogenic when injected into sublethally-irradiated NOD/SCID mice (Akimov et al., 2005). These findings establish the feasibility of bypassing senescence in human hematopoietic progenitors through genetic engineering, providing proof-of-principle for approaches that might eventually lead to the establishment of permanent human HSC lines. Accordingly, our future efforts will focus on extending these results by assessing the combinatorial effects of novel hematopoietic growth factors such as Notch ligands, Hedgehog proteins, Wnt molecules, bone morphogenic proteins, HOXB4 homeoprotein, and angiopoietin-like proteins (Sauvageau et al., 2004; Zhang et al., 2006a).

Table 1.

Characteristics of hTERT- plus HPV16 E6/E7-immortalized human cord blood-derived hematopoietic progenitor cell lines

Cell Line	ET1a	ET2
Cell Surface Phenotypea	CD133loCD235aloCD71+CD203c+CD33+CD13+	CD133loCD235aloCD71+CD203c+CD33+CD13+
Growth factor responsivenessb	SCF-dependent	SCF-dependent
Karyotypec	46,XY,der(22)t(17;22)[9]	45,XY,der(14)t(9;14),der(19)t(19;22),-22[8]

^aIn addition to the CD133 cell surface antigen, candidate human HSCs have been suggested to express CD33 and CD13 (Taussig et al., 2005).

^bThe ET1a and ET2 human hematopoietic progenitor cell lines require SCF for survival and proliferation but grow optimally in the presence of SCF, TPO, Flt3 ligand and IL-3. Based on growth factor responsiveness, the cell lines are presumed to express CD117 (c-Kit receptor) and CD123 (the low affinity binding subunit of the IL-3 receptor).

^cSee Akimov et al. (2005) for details.

Derivation of Human Hematopoietic Progenitor Cell Lines Protocol

1. The HIV-1-based SINF-MU3-hTERT-IRES-GFP-W-S and SINF-MU3-E6E7-IRES-YFP-W-S lentiviral vectors used to immortalize human hematopoietic progenitors have been described previously (Akimov et al., 2005). SINF-MU3-hTERT-IRES-GFP-W-S contains the hTERT cDNA upstream of an encephalomyocarditis virus internal ribosome entry site (IRES)-green fluorescent protein (GFP) gene cassette and SINF-MU3-E6E7-IRES-YFP-W-S contains the HPV16 E6/E7 coding region upstream of an IRES-yellow fluorescent protein (YFP) gene cassette. In both cases, transgene transcription is driven by an internal murine stem cell virus (MSCV) long terminal repeat (LTR) promoter (Hawley et al., 1994; Ramezani et al., 2000; Ramezani et al., 2003).

2. Culture human embryonic kidney 293T cells in Dulbecco’s modified Eagle’s medium (Invitrogen) supplemented with 10% heat-inactivated FBS, 2 mM L-glutamine, 50 IU/ml penicillin and 50 μg/ml streptomycin at 37° in a humidified atmosphere containing 5% CO₂. Plate the 293T cells (4 × 10⁶) into 10-cm tissue culture dishes containing 7 ml of complete medium the day before transfection. Mix 15 μg of the transfer vector plasmid (SINF-MU3-hTERT-IRES-GFP-W-S or SINF-MU3-E6E7-IRES-YFP-W-S), 10 μg of the packaging plasmid pCMVΔR8.91 (Zufferey et al., 1997) and 5 μg of the VSV-G glycoprotein envelope plasmid pMD.G (Naldini et al., 1996). Bring the volume up to 450 μl with sterile water. Add 50 μl 2.5 M CaCl₂ and mix. Add the DNA/CaCl₂ solution dropwise to 500 μl of 2 × N-(2-Hydroxyethyl)piperazine-N′-2-ethanesulfonic acid (HEPES)-buffered saline (0.283 M NaCl, 0.023 M HEPES [Sigma-Aldrich, Catalog number H0887], 1.5 mM Na₂HPO₄, pH 7.05) in a 15-ml conical tube. Use a 5 ml pipet to bubble the 2 × HEPES-buffered saline while adding the DNA/CaCl₂ solution. Vortex immediately for 5 seconds and incubate for 20 min at 22°. Add the precipitate dropwise over the cells and mix gently. Incubate the cells overnight (16 hours) at 37°. The next day, remove medium from the plate, rinse cells with 5 ml PBS, and add 7 ml fresh medium. Collect the vector-containing medium after another 48 hours, centrifuge at 2000 g for 10 min to remove cellular debris and filter through a 0.45-μm pore-size filter (Nalgene). Ultracentrifuge vector supernatants in 70-ml bottles (Beckman Coulter) at 45,000 g for 90 min at 4°. Resuspend pellets in 500 μl medium by gentle vortexing for 2 hours at 4°. Spin down the debris at 2000 g for 5 min and store the concentrated vector particles at −80°. Titer vector stocks on human fibrosarcoma HT1080 cells and assay for the presence of replication-competent virus as previously described (Ramezani and Hawley, 2002b; Ramezani and Hawley, 2003).

3. Coat 24-well non-tissue culture-treated plates (Lux Suspension Dish, Fisher Scientific, Catalog number ICNLX171099) with 2 μg/cm² recombinant fibronectin fragment (RetroNectin, Takara Mirus Bio, Catalog number TAK_T100A). Culture CD34⁺ cells isolated as described in the previous section at a density of 1 × 10⁶ cells/ml for 24 hours in X-VIVO-15 serum-free medium (Fisher Scientific, Catalog number BW04-418Q) supplemented with 10% BIT 9500 serum substitute (bovine serum albumin, insulin, and human transferrin; StemCell Technologies, Catalog number 09500), 100 μM β-mercaptoethanol, 100 ng/ml SCF, 20 ng/ml TPO and 100 ng/ml Flt3 ligand, with or without 20 ng/ml IL-3 (all cytokines from PeproTech) at 37° in a humidified atmosphere containing 5% CO₂. Transduce the cells with lentiviral vector particles (2 × 10⁶ transducing units/ml; multiplicity of infection, 2) in the presence of 4 μg/ml protamine sulfate (Sigma-Aldrich) (Ramezani et al., 2003; Ramezani and Hawley, 2002b; Ramezani and Hawley, 2003). Change the medium after 24 hours and continue culturing the cells.

4. Harvest the hematopoietic progenitor cells after an additional 48–72 hours of culture (Cell Dissociation Buffer, Invitrogen, Catalog number 13151-014), wash and resuspend in PBS containing 2% FBS. Under the conditions employed, the majority of the cells retain the CD34⁺ phenotype (Ramezani et al., 2000). Isolate GFP⁺YFP⁺ cells to >95% purity by fluorescence-activated cell sorting (Akimov et al., 2005; Cheng et al., 1997; Dorrell et al., 2000; Hawley et al., 2004). The YFP and GFP signals are separated with a 525 nm shortpass dichroic filter and collected with a 550/30 nm BP filter and a 510/20 nm BP filter, respectively (Omega Optical, Catalog number XCY-500).

5. Maintain the hematopoietic progenitor cells in continuous culture in X-VIVO-15 serum-free medium supplemented with 10% BIT 9500 serum substitute, 100 μM β-mercaptoethanol, 100 ng/ml SCF, 20 ng/ml TPO and 100 ng/ml Flt3 ligand, with or without 20 ng/ml IL-3 at 37° in a humidified atmosphere containing 5% CO₂.