¹Department of Medicine, University of Pennsylvania School of Medicine, Philadelphia, Pennsylvania, PA 19104, USA

²Department of Pediatrics, University of Pennsylvania School of Medicine, Philadelphia, Pennsylvania, PA 19104, USA

Correspondence to: S G Emerson, Division of Hematology/Oncology, Departments of Medicine and Pediatrics, Maloney 510, 3600 Spruce Street, University of Pennsylvania School of Medicine, Philadelphia, PA 19104, USA; E-mail: emersons@mail.med.upenn.edu

The past two decades have witnessed significant advances in our understanding of the cellular physiology and molecular regulation of hematopoiesis. At the heart of stem cell self-renewal and lineage commitment decisions lies the relative expression levels of lineage-specific transcription factors. The expression of these transcription factors in early stem cells may be promiscuous and fluctuate, but ultimately comes under the influence of extracellular regulatory signals in the form of hematopoietic cytokines. In this review, we first summarize our current understanding of the phenotypic characterization of hematopoietic stem cells. Next, we describe key known transcription factors which govern stem cell self-renewal and lineage commitment decisions. Finally, we review data concerning the role of specific cytokines in influencing these decisions. From this review, a picture emerges in which stem cell fate decisions are governed by the integrated effects of intrinsic transcription factors and external signaling pathways initiated by regulatory cytokines.

Oncogene (2002) 21, 3295-3313 DOI: 10.1038/sj/onc/1205318

transcription factor; cytokines; hematopoiesis

The cell biology of primitive stem cells

Diverse results from many experimental systems indicate that the proliferative and self-renewal capacity of a given hematopoietic cell, as well as the diversity of lineages to which it can differentiate, decrease progressively with differentiation (Figure 1). In the now classical view, the earliest cell in this hierarchy is the pluripotent hematopoietic stem cell (HSC), whose existence was demonstrated by serial bone marrow transplantation (BMT) experiments. In these studies, all lineages of derived hematopoiesis could be successfully reconstituted in lethally irradiated tertiary recipient mice by bone marrow cells from secondary recipients. Furthermore, successful reconstitution could also be achieved by using many fewer stem cell-enriched cells, isolated by any one of a variety of methods, either from bone marrow, peripheral blood or fetal liver. Most recently, careful studies suggest that within the subset of cells previously viewed as uniquely hematopoietic stem cells may lie a subset with truly multi-lineage hematopoietic potential, capable of giving rise to many non-hematopoietic lineages as well, including myocytes, epithelial cells, hepatocytes, and neurons.

The phenotypic specification has become progressively more sophisticated over the past two decades. Although the precise relationship between cell surface phenotype and cellular plasticity within the stem cell compartment is still not clear, the murine Sca-1⁺ C-kit⁺ Lin^- (SKL) population clearly contain most, if not all of the cells with HSC potential, and the same is likely true of the equivalent population in humans. This SKL population is further characterized by the differential expression of CD34 and CD38, with SKL cells being functionally divided between CD34⁺ and CD34^-/lo sub-populations. The CD34⁺ cells provided immediate radioprotection for lethally irradiated recipient mice while CD34^-/lo cells were responsible for long-term reconstitution of both myeloid and lymphoid lineages (Osawa et al., 1996). The authors postulated that CD34⁺ sub-population basically consists of multilineage progenitor cells with low self-renewal capacity, while the real stem cells are contained in CD34^-/lo fraction. Similarly, it was recently reported that only the descendants from CD38⁺CD34^- SKL cells, but not from CD34⁺ SKL cells could be easily detected in the re-constituted BM cells in lethally irradiated secondary, and tertiary recipients over a long term (Zhao et al., 2000).

Developmentally, there are two waves of hematopoiesis, so probably with two corresponding origins for HSCs (Ciau-Uitz et al., 2000). Primitive hematopoiesis, distinguished from definitive hematopoiesis by large and nucleated erythrocytes and specifically expressed fetal hemoglobin isoforms, occurs as a transient wave preceding the advent of definitive hematopoiesis. Primitive hematopoiesis in the mouse begins within the blood islands in yolk sac around 7.5 days post coitum (d.p.c.). Independently, the stem cells for definitive hematopoiesis originate within embryonic SP/AGM (splanchnopleur, aorta, gonads, and mesonephros) during 9.5 to 11.5 d.p.c. in mouse and 30 to 37 days of gestation in the human, where the hematopoietic cells are found adhering to the ventral wall of dorsal aorta (Medvinsky and Dzierzak, 1996). Only these definitive hematopoietic stem/progenitor cells originating from AGM, but not primitive HSCs' are able to repopulate the entire hematopoietic system in the lethally irradiated adult recipient mice (Cumano and Godin, 2001). Definitive HSCs expand in number in the AGM, and then migrate to and colonize fetal liver and spleen where they continue to differentiate into recognizable hematopoietic precursors. After birth, definitive hematopoiesis is primarily confined to bone marrow, and in some pathological conditions also to extramedullary sites such as spleen, liver, and occasionally lung and brain.

Because both endothelial and hematopoietic cells seemed to simultaneously derive from clusters of phenotypically similar cells within the yolk sac, it has been suggested that hematopoietic and endothelial cells share an immediate embryonic parental cell, either a hemangioblast that serves as the common precursor for hematopoietic and endothelial lineages, or a hemogenic endothelium that gives rise to hematopoietic stem cells (see Figure 1). Although definitive hematopoietic stem/progenitor cells do not resemble endothelial cells morphologically, they do share a number of transcription factors and surface markers, including SCL, GATA-2, C-kit, AA-4.1, CD34, Flit-3 ligand, Sca-1, VEGFR-1 and -2, only with the exception of CD45. Actually endothelial specific receptor VEGFR-2 was recently found a key marker for isolating hematopoietic reconstituting cell in NOD/SCID mice from human circulating CD34⁺ cells (Ziegler et al., 1999). This close relationship between HSC and endothelial development can be further visualized in embryonic stem cell cultures. During the in vitro embryo body development of mouse, the blast colony-forming cells (BL-CFC) develop within 4 days in the presence of vascular endothelial growth factor (VEGF) and are lost quickly. These BL-CFCs within embryonic bodies can produce adherent endothelial cells, and primitive or definitive erythroid cells, as well as macrophages and neutrophils recognized by morphology examination. These BL-CFCs may thus indeed be the in vitro counterpart to the long-hypothesized hemangioblasts (Choi et al., 1998). It is also possible that hemangioblasts may persist after birth. Shi et al. (1998) showed evidence indicating that the endothelial cells lining the Dacron graft implanted in the aorta were actually derived from the pre-transplanted BM cells 8 weeks before. They also showed that some of CD34⁺ BM cells could be induced to endothelial cells in vitro culture system with cytokines like VEGF.

One of the most exciting findings in the past 2-3 years has been that cells with the phenotype of hematopoietic stem cells, but derived from non-hematopoietic tissues such as brain and muscle, can efficiently repopulate hematopoiesis in lethally-irradiated mice (Bjornson et al., 1999; Jackson et al., 1999). Conversely, numerous laboratories have found that transplanted BM cells could contribute to the regeneration of multiple tissue cell types, including brain, muscle, hepatocytes, lung, GI epithelium, and skin (Brazelton et al., 2000; Gussoni et al., 1999; Krause et al., 2001; Lagasse et al., 2000; Mezey et al., 2000). These two sorts of findings have given rise to two related hypotheses: (1) that HSCs are ubiquitous in mature mammals, and the restricted appearance of hematopoietic differentiation in bone marrow alone is due to environmental regulatory controls on HSC differentiation; and (2) that HSCs are only a subset, either phenotypically or transcriptionally defined, of a pool of ubiquitous multi-tissue stem cells, perhaps closely related to embryonic stem cells.

Transcriptional regulation of early hematopoietic development

Since purifying and biochemically interrogating very rare HSC populations is so difficult, the role of specific transcription factors in HSC fate decisions has derived largely from genetic strategies, primarily gene-targeting (knockout) and retroviral infection/overexpression experiments. From this growing body of literature, several transcription factors have been found to play critical roles in HSC physiology, including SCL (stem cell leukemia hematopoietic transcription factor), GATA-2, and Lmo-2, which are essential for primitive and definitive hematopoiesis, and AML-1 that is required for definitive hematopoiesis.

SCL/Tal

SCL homozygous null embryos died at around 9.5 d.p.c., and histological studies showed complete absence of recognizable hematopoiesis in yolk sac, with GATA-1 and c-Myb expressions being undetectable (Robb et al., 1995). SCL null ES cells were not able to form any lineages of blood cells in chimeric mice. SCL null ES cells were also unable to form blast colonies in vitro, but only transitional colonies that contained only low numbers of primitive erythroid precursors and a subset of precursors associated with early stage of definitive hematopoiesis (Robertson et al., 2000). In control and SCL-null EB, the temporal expression pattern of genes associated with the formation of ventral mesoderm, such as Brachyury, BMP-4 (bone morphogenetic protein-4), and Flik-1 (VEGF receptor) was identical. The GATA-2, CD34, and C-kit that are co-expressed in endothelial and hematopoietic lineages were expressed normally in SCL-null embryonic stem cell lines. However, hematopoietic-restricted genes, including transcription factors GATA-1, EKLF (erythroid kruppel-like factor), and PU.1 as well as globin genes and MPO (myeloperoxidase), were only expressed in wild type and SCL-heterozygous ES cells. SCL protein shares homology within a restricted region of 56 amino acids with a number of bHLH transcription factors, and it is capable of binding to E-box motif of DNA (CANNTG) in vitro, and forms heterodimers with other HLH factors, such as E47 and E12. Strikingly, SCL mutants unable to bind to E-box were found to be able to nearly fully rescue the hematopoietic defect of SCL homogeneous null Zebra fish (Porcher et al., 1999). The studies' authors proposed two models to explain this phenomenon: (1) the primary role of SCL here is to sequester an unidentified repressor, for which DNA-binding of SCL is dispensable; (2) this deficiency in DNA binding ability of SCL could be compensated by the affinity binding with its transcriptional partners retaining an intact DNA binding domain to target genes' promoter.

GATA-2

Both primitive hematopoiesis in yolk sac and definitive hematopoiesis in fetal liver and spleen as well as adult bone marrow are greatly affected by the deficiency of GATA-2. Deficient primitive hematopoiesis, evident by severe anemia, leads most GATA-2-deficient embryos to die before 10.5 p.c. When GATA-2 ^-/- ES cells were injected into wild type blastocysts, the contribution of GATA-2 ^-/- cells in both fetal liver and adult hematopoietic compartment of chimeric mice were nearly undetectable (Tsai et al., 1994). The introduction into deficient ES cells with a 250 kb GATA-2 YAC clone, which was strongly expressed in both primitive and definitive compartments, rescued hematopoiesis both in vivo and in vitro (Zhou et al., 1998). In patients with aplastic anemia, the GATA-2 expression was found decreased in CD34⁺ bone marrow cells, while no significant change in SCL and AML1 expression was detected (Fujmaki et al., 2001). Both wild type PML and PML-RAR fusion proteins were able to interact with GATA-2 and potentiate its transcription activity, implicating a role of GATA-2 in the molecular pathogenesis of APL (Tsuzuki et al., 2000). GATA-2 may also be one major mediator of the ability of BMP-4 to specify hematopoietic mesoderm, since BMP-4 expression in the embryo is polarized to the ventral wall of the aorta, immediately underlying the site of initial hematopoiesis (Marshall et al., 2000; Maeno et al., 1996) (see Figure 2).

LMO2

Lim-finger protein LMO2, which was found to be activated in T cell leukemia by chromosomal translocation, is likewise indispensable for primitive erythropoiesis and definitive hematopoiesis, as shown in a Lmo2 null/wild type chimeric mice (Yamada et al., 1998). Embryogenic Lmo2 expression was found localized to hematopoietic sites (Manaia et al., 2000).

AML-1

Homozygous knockout mice showed absent fetal liver hematopoiesis, and failed to contribute to definitive hematopoiesis in chimeric mice, but primitive yolk sac erythropoiesis was unaffected (Okuda et al., 1996). The targeted disruption of its partner-CBF similarly results in a complete block of fetal liver hematopoiesis. The rescue of AML-1 deficient ES cells by knocking-in of a wild type AML-1b restored their ability to contribute to the formation of every lineage of hematopoietic cells in chimeric mice (Okuda et al., 2000). In vitro rescue experiment of the deficient ES cells by the retroviral vectors carrying a series of AML1b mutants indicated that a 61aa of C-terminal region containing a VWRPY motif, required for interacting with transcriptional co-repressors, was not required for the definitive hematopoiesis (Okuda et al., 2000). Taken together, these results suggest that the transcriptional activation, rather than repression of target genes by AML-1 is required for definitive hematopoiesis.

Regulation of self-renewal and differentiation of HSCs

The transcriptional machinery governing primitive stem cell biology is undoubtedly very complex. While the transcription factors described above are absolutely essential for the survival and proliferation of HSCs, other molecules clearly impact strongly on the cell fate decisions of stem cells, both for symmetric expansion and lineage commitment. While we are still at the beginning of uncovering these key molecular regulators, work indicates that transcription factors HoxB4 and Ikaros, activated nuclear form of Notch1, cell cycle inhibitor P21, and TGF/BMP-4 family members as well as TNF receptor P55 signaling are likely involved in the maintenance or promotion of the hematopoietic stem cell renewal (see Figures 1 and 2).

Regulation by transcription factors

Homeobox genes

Several members of the homeobox gene family are expressed differentially during hematopoieic differentiation. Perhaps because of the close similarity and overlapping functional roles of these molecules, the results of gene knockout studies have been modest. Overexpression of Hox genes by stem cell retroviral transduction, in contrast, have been very revealing. For examples, it was first noticed that in vitro differentiation of the human Lin^-, non-adherent peripheral blood cells to either erythroid or granulocytic direction was accompanied by a remarkable and persistent increase in the HoxB4 mRNA level, and that the administration of HoxB4 antisense oligo inhibited the colony formation to these two lineages in vitro (Giampaolo et al., 1994). Then it was shown that long-term repopulating ability of murine bone marrow cells was increased to at least 10-fold when HoxB4 cDNA was over-expressed by retroviral infection, in comparison to control stem cells infected with empty vector (Sauvageau et al., 1995; Thorsteinsdottir et al., 1999). Paradoxically, it seemed the CRU (competitive repopulating unit) capacity of bone marrow cells from the HoxB4 over-expression group never exceeded the CRU limit found in the normal mice. It was also reported by the same lab that the over-expression of HoxB4 in ES cells prompted the production of definitive GMME-CFU and CFU-E colony in the in vitro differentiation model (Helgason et al., 1996). Recently, HoxB4 expression was found increased in protein level in the myeloid differentiation of HL-60 cells induced by VitD3. By delivering Hox B4 antisense oligonucleotides into HL-60 cells, authors showed that the monocytic differentiation was inhibited (Pan and Simpson, 2001). These effects may be mediated through HoxB4 proteins interactions with the c-Myc gene and the cell cycle machinery (Antonchuk et al., 2001). HoxB4 protein was identified as the transcription elongation-blocking factor binding to MIE1 site within the intron1 of c-Myc gene (Pan and Simpson, 1999). In the non-hematopoietic polyclonal Rat-1 fibroblast cells, the over-expression of Hox B4 resulted in the faster growth, reduced requirement to serum, and even malignant transformation, in cooperation with PBX1. This phenomenon was accompanied by enhanced AP-1 activity and increased level of cyclin D1 (Krosl et al., 1998). It is also interesting to note that the overexpression of the HoxB4 paralogue HoxC4 in primitive hematopoietic cells also enhances the proliferation of both HSCs and committed progenitors, although the HoxC4's effect may be milder than HoxB4's (Daga et al., 2000).

If HoxB4 gene products are critical regulators of HSC cycling, then the molecular mechanism governing HoxB4 expression is clearly critically important to primitive stem cell fate decisions. Recently, the two essential DNA binding sites were identified in the promoter region of HoxB4 gene functioning in both normal and malignant hematopoietic primitive cells, just upstream of its transcriptional starting site. One of them was a classical E box and served as the functional binding site for USF1 and USF2 (Giannola et al., 2000).

Ikaros

Mice lacking all isoforms of Ikaros display decreased expressed of Flk-2 and C-kit on their SKL cells, and show a marked reduction in long-term repopulating units as measured by competitive repopulation, and mice homozygous for Ikaros dominant negative mutation (DNA-binding domain mutated) possess no measurable repopulating activity at all (Nichogiannopoulou et al., 1999). The Ikaros gene products belong to zinc finger family of transcription factors. Long form of Ikaros mainly consists of two zinc fingers, an N-terminal finger domain mediating DNA binding and C-terminal domain mediating dimerization. Generally, Ikaros proteins modulate transcription by recruiting co-repressor complex to the promoters of target genes and/or sequestering these genes to the vicinity of hetero-chromatin (Koipally et al., 1999; Sabbattini et al., 2001; Trinh et al., 2001). Intriguingly, different isoforms of Ikaros are differentially expressed at different stages of hematopoiesis (Klug et al., 1998), suggesting that regulated expression of Ikaros isoforms could provide fine regulation of the expression of lineage specific transcription factors.

Notch1

In contrast to the transcription factors described above, which are believed to be directly regulated by only intracellular events, the Notch1-Jagged pathway may provide a key pathway to integrate extracellular regulatory signals with stem cell cycling control. Following engagement of the Notch1 receptor by extracellular ligand, cleavage events release the intracellular portion of Notch1, which in turn translocates to the nucleus and acts as a transcription factor on its target genes. By constitutively over-expressing the intracellular domain of Notch1, several immortalized cell lines were established from murine BM SKL cells, which still maintained the potential to reconstitute myeloid and lymphoid cell lines both in vivo and in vitro (Varnum-Finney et al., 2000). This phenomenon may reflect the in vivo physiology of Notch/jagged interactions, as the addition of Notch ligand jagged-1 into in vitro expansion system of human CD34⁺CD38^- Lin^- cord blood cells prompted their ability to repopulate pluripotently in SCID mice (Karanu et al., 2000).

Most importantly, the list of transcription factors and cell cycle regulatory molecules studied to date may represent only a small subset of the genes that control stem cell self-renewal and lineage commitment decisions. Other candidate genes critical for stem cell regulation may surface from broader screening of genes differentially expressed in HSCs versus closely related but non-HSC cells. For example, by comparing a cDNA library of murine fetal liver HSCs (SKL AA4.1⁺) with that made from stem cell-depleted AA4.1^- cells of same tissue, investigators have identified a large number of genes that are preferentially expressed in stem cells versus more differentiated cells. The known transcription factors or chromatin binding proteins on the list include AML-1, ALL-1, Bnmt-3b, Evi-1 and macroH2A1.2 and they may involve the maintenance of gene expression program unique to HSCs (Philips et al., 2000).

Regulation by stromal cells and cytokines

In many biological systems it has been shown that milieus where the stem cells reside play critical roles in the determination of their self-renewal or differentiation choices. For example, in Drosophila the germ line stem cell renewal/differentiation balance is maintained by surrounding somatic cells that direct the asymmetric division of stem cells. The maintenance of this precise balance involves activation through the epidermal growth factor receptor in somatic cells (Kiger et al., 2000). In the case of murine spermatogonial stem cells, glial cell line-derived neutrophic factor, a distant member of transforming growth factor-1 family secreted by Sertoli cells, was found to maintain the stem cell reserves and inhibit their differentiation (Meng et al., 2000). Likewise, murine ES cells can be successfully maintained in an undifferentiated state and propagated in vitro by the presence of leukemia inhibitory factor (LIF) (Ichikawa, 1970), which requires the activation of downstream molecule Stat3 in ES cells (Raz et al., 1999; Niwa et al., 1998). LIF is normally produced in embryonic fibroblasts adjacent to ES cells in vivo.

Adjacent cells and local cytokines may also regulate self-renewal versus differentiation decisions faced by HSCs. For example, one group of investigators established two endothelial cell lines from day 11 murine CD34⁺ AGM cells by infecting them with retroviruses over-expressing polyoma middle T antigen. One of these immortalized stromal lines supported the expansion of CD34⁺ SKL fetal liver cells, while a second line induced the differentiation of these cells into erythroid, myeloid and B lymphoid cells. For the self-renewal induction, the direct contact of hematopoietic cells with these established endothelia was required (Ohneda et al., 1998). Circumstantial evidence suggests that similar local interactions between specific stromal cells and HSCs may exist in vivo. For example, the ability of bone marrow stromal cells to support hematopoiesis was found to increase between day 7 and 10 after birth, correlating with the emergence of hematons in the bone marrow, the compact cellular aggregates consisting of various stromal cells and hematopoietic cells. When studied functionally, this hematon fraction was found 3.7-fold enriched in day 35 LTC-IC over unfractionated non-adherent buffy coat bone marrow cells (Blazsek et al., 2000).

Evidence from several systems suggest that the most primitive repopulating HSCs are largely quiescent, with the majority of these cells out of cycle at any one time. It is well known that low level of TGF-1 maintain the multiple differentiation potentials of hematopoietic stem/progenitor cells, which is associated with its negative controlling effect on cells cycling (Batard et al., 2000). Accordingly, in another study, TGF-1 was found to slow down the growth of human umbilical cord CD34⁺CD38^-Lin^- blood cells in vitro, accompanied by the up-modulation of cell cycle inhibitor p21, while independent of its downregulating effect on Bcl-2 expression (Ducos et al., 2000; Francis et al., 2000). It was recently suggested that the G1 checkpoint regulator p21 is one key factor required for the maintenance of this quiescent status, for BM cells homozygous null for p21 failed to long-term repopulate the hematopoietic system of lethally irradiated mice. These data were interpreted to suggest that failure to maintain stem cells out of cell cycle might predispose stem cells to rapid exhaust after transplantation (Cheng et al., 2000b). In the contrast, this same experimental system was exploited to show that knockout of another cyclin-dependent kinase inhibitor-p27 prompted the repopulating efficiency of BM cells, possibly through enhancement of proliferating ability of progenitor cells, and with no disturbance on the stem cell quiescence (Cheng et al., 2000a). However, upon the stimulation of leukemia cell line Kasumi-1 by TGF-1, the leukemia fusion protein AML1-ETO can sequester Smads, which are the essential downstream effectors of TGF-1 signaling pathway, so as to inhibit the expression activation of Smad target genes, suggesting a quiescence-independent mechanism for maintaining leukemia stem cell renewal (Jakubowiak et al., 2000).

There are several other interesting candidates for the cytokines that might mediate these stem cell maintenance, self-renewal and expansion in vivo. Early experiments indicated that IL-3 and GM-CSF might be involved, as IL-3 sustained the renewal of FDCP cell line while GM-CSF induced them to macrophage-granulocyte differentiation (Just et al., 1991). However, these molecules are unlikely to play essential roles in vivo, at least by themselves. IL-3 is not expressed in bone marrow stroma. On the other hand, the combination of SCF, Flt-3L and Tpo with either IL-11 or IL-3, was able to support in vitro expansion of long term multilineage repopulating activity of SKL CD34^- cells for at least five cell divisions (Bryder and Jacobsen, 2000). More recently, newer candidates for soluble molecules known to be present on stromal cells have arisen as stem cell proliferation promptors. Tie-2, a tyrosine kinase receptor for angiopoietin, on the fetal liver KLSA (AA4.1⁺) cell population was found in association with the cells with long term multiple-lineage repopulating activity (Hsu et al., 2000). The soluble form of sonic hedgehog, along with human homologues to its receptors, was found to be expressed in primitive and mature blood cells. These findings are very provocative, since high concentration of sonic hedgehog protein were found to expand human CD38^-34⁺ Lin^- cells in vitro, which was dependent on the transcription activation of BMP-4 (Bhardwaj et al., 2001), and polarized expression of BMP-4 has been found in endothelium immediately underneath definitive hematopoietic clusters in the embryo. Furthermore, in vitro study shows that low doses of BMP-4 promoted the proliferation and differentiation of human CD38^-Lin^- cells, while higher concentration of BMP-4 extended the time from 4 to 6 days until that the repopulating capacity of cultivated cells could be maintained (Bhatia et al., 1999).

A possible role for TNF in the regulation of HSC proliferation seems quite possible, however the data concerning its effect on HSCs are conflicting. On one hand, TNF was found to eliminate long-term culture-initiating cells from human CD34⁺CD38^- bone marrow, via signaling through the ceramide pathway, at a concentration lower than what is required for initiating apoptosis (Maguer-Satta et al., 2000). In another study, TNF, through its binding to the P55 receptor, inhibited the maintenance and expansion of multipotent repopulating ability of CD34⁺CD38^- cells cultured in vitro in the presence of SCF/FL/TPO, independent of apoptosis induction and cell cycle change. This TNF effect was postulated to reflect a biased effect on differentiation over renewal of HSCs, especially to an accelerated myeloid cell commitment and turnover (Dybedal et al., 2001). In direct contrast, however, the P55 TNF receptor knockout mice were found to have impaired HSC repopulating, despite the overt phenotype of increased BM cell cellularity, suggesting that a dosage-dependent effect of TNF on HSC renewal versus differentiation (Rebel et al., 1999).

Early stem cell decisions: commitment to myeloid and lymphoid lineages

Once HSCs divide and generate more differentiated daughter cells, within 10-15 divisions the genetic programs of the descendent cells become fixed toward a single lineage. The first steps in this process of lineage restriction are uncertain, but the dominant hypothesis today proposes that the first decisions involve restriction to the ability to generate myeloid versus lymphoid progeny. The first cells generated following such decisions have been termed common lymphoid precursors (CLPs) and common myeloid precursor (CMPs) (Figure 1). Kondo et al. exploited the data that IL-7 receptor is widely expressed on mature lymphoid (Peschon et al., 1994), but not myeloid cells, to isolate a potential CLP population with the phenotype IL-7R⁺Lin^-Sca-1^loC-kit^lo from adult murine BM. This same population was later proven to possess short-term (4-6 weeks) repopulating ability restricted to either B, T or NK cells in vivo (Kondo et al., 1997). These cells, which comprise 0.02% of BM cells would appear to meet the criteria for CLPs (Kondo et al., 1997). Of note, GATA-3 and Aiolos were expressed while the expression of GATA-1, C/EBP and NFE2 that are thought to associate with myeloid development were repressed (Akashi et al., 2000b).

Committed myeloid precursors, in contrast, were isolated from IL-7R^-/Lin^-C-kit⁺/Sca-1^- populations. This subset was further divided into three sub-populations by the differential expression of CD34 and FcR: (1) CD34⁺ FcR^lo, (2) CD34^-FcR^lo and (3) CD34⁺FcR⁺ (Akashi et al., 2000b), and the potential of each sub-population was examined by in vitro methylcellulose CFU culture and in vivo BMT assays. Of them, the CD34⁺ FcR^lo cells constitute about 0.2% of BM cells and gave rise to daughter cells consisting of monocytes, granulocytes, erythrocytes and megakaryocytes, while CD34^-FcR^lo cells and CD34⁺FcR⁺ cells exclusively produced megakaryocytes/erythrocytes and granulocytes/monocytes, respectively. The CD34⁺FcR^lo cells, like CLPs, only capable of short-term (2 weeks) repopulating the lethal irradiated mice, are considered as the common myeloid precursors (CMP). It was also found that the short-term cultivation of CMP cells generated CD34^-FcR^lo (megakaryocytic/erythroid precursor, MEP) and CD34⁺FcR⁺ (granulocytic/monocytic precursor, GMP) cells as thought of before (see Figure 3). The transcription factors SCL, GATA-2, NF-E2, GATA-1, C/EBP, c-Myb, and PU.1, and the cytokine receptors TpoR and EpoR, were found to be expressed in CMP (Akashi et al., 2000b).

Additional experiments suggest that production of monocytes may be more promiscuous, and occurs from common lymphoid as well as myeloid precursors (Kee and Murre, 1998) (see Figure 4). Clonogenic fetal liver cells with the phenotype of AA4.1⁺B220^-Mac-1^-Sca-1⁺ were first shown to contain the bi-potent precursor for B cells and macrophages (Cumano and Godin, 2001). In a second study, AA4.1FcRII/III⁺ cells from the murine fetal liver of 13 dpc were found to be precursors possessing potential to produce B cells, T cells and macrophages (Lacaud et al., 1998). And more recently, a postnatal mouse bone marrow cell population with phenotype IL-7R⁺Sca-I^-C-kit^-Lin^-B220^-CD19⁺ was isolated, constituting about 0.5% BM cells (Montecino-Rodriguez et al., 2001), which produced only B cells or macrophages in vitro.

How are lineage commitment decisions from HSCs initiated and regulated?

This critical but most difficult question has been the subject of intense and creative debate, and equally intense and creative experimentation. Among the most intriguing set of relevant data are those suggesting that HSC, CLP and CMP fate decisions might not be instantly made, but rather preceded by a phase of promiscuousness or hesitation (Enver and Greaves, 1998; Rothenberg, 2000). For example, the mature erythroid differentiation marker -globin and monocytic marker MPO were unexpectedly found to be co-expressed, at low levels, in the single cell of multiple-potent FDCP-Mix4 cells (Hu et al., 1997). In a transgenic mouse model, an erythroid specific, micro-LCR-hemoglobin promoter-coupled Lac-Z was not only found to express in erythroid lineage, but also in megakaryotic, and in monocytic/granulocytic progenitors in vivo, though to milder extents (Papayannopoulou et al., 2000). These data indicate that the alternative differentiation potential might not be immediately eliminated, but rather repressed in a graded or gradual way in the cells that are committing to a given lineage.

Presumably, the differential expression of transcription factors triggers the determination of HSC fate: renewal, or commitment to either CLP or CMP. By using differential cDNA library and sensitive RT-PCR method, a number of candidate transcription factors that are preferentially expressed in cells destined to one of those fates have been found. However, most of these transcription factors have not been tested in clearly interpretable in vivo models. Moreover, the primary events and mechanism leading to the induction of differentially expressed genetic programs are obscure. Of the transcription factors studied to date, the largest body of published evidence relates to the transcription factor PU.1. Predominance of PU.1 could be among one of earliest events biasing HSCs to lineage commitment, for the co-upregulation of PU.1 and GATA-1 heralds the commitment to CMPs, and PU.1 expression is maintained in CLPs and also absolutely required for lymphoid development, while knockout of PU.1 spared the development of primitive erythropoiesis (Scott et al., 1994). PU.1 involves the transcription regulation of both IL-7 receptor and M-CSF receptor, which are required for proper developments of lymphoid and myeloid lineages, respectively (DeKoter and Singh, 2000; DeKoter et al., 1998). PU.1 over-expression alone was shown to be able to commit multipotent hematopoietic progenitors to monocytic/granulocytic direction (Nerlov and Graf, 1998). Two clues have been found to account for how the active function of PU.1 could lead to two mutually exclusive outcomes: lymphoid versus myeloid cells. Firstly, the distinct effects of PU.1 on lineage commitment is dosage-related, with a higher concentration favoring HSCs commitment to myeloid cells (DeKoter and Singh, 2000). Secondly, the PU.1 activity on its target gene could be negatively regulated by the lymphoid cell transcription factor Pax5 (Maitra and Atchison, 2000; Nutt et al., 1999), which is preferentially expressed in B lymphocytes and absolutely required for B lymphocyte commitment.

Data is scarcely available in terms of the molecular mechanisms regulating the expression of key transcription factors like PU.1, and of how the expression of differentiation-inducing transcription factors might predominate over self-renewal-maintaining factors. In one view, the so-called 'intrinsic theory,' a transcription factor expression pattern is initiated and even reached by the stochastic and autonomous induction of transcription factors. Some evidence in support of this view that has been presented involves the stochastic transcription activation of transcription factor Pax5 (Nutt and Busslinger, 1999). Whether or not such intrinsic transcription factor induction may occur, the stable expression of transcriptional programs might still be subject to instruction from outside signals, as proclaimed by the 'extrinsic theory'. The latter was recently supported by experiments in which the addition of anti-TGF-1 antibody into in vitro culture system of human CLRPP (cytokine low-responding proliferating progenitors) cells resulted in an increased generation of cells expressing CD15/CD11b/glycophorin-A, accompanied by a significant upregulation of PU.1 and GATA-1 expression level (Pierelli et al., 2000). Thus, at least in this in vitro setting, TGF-1 would appear to directly inhibit the commitment of HSCs to CMP.

Related evidence is accumulating for the potential of extrinsic cytokines to positively direct lineage commitment decisions. Based on the finding that low but detectable levels of GM-CSF receptor were found to be expressed on HSCs but not CLPs, it was suggested that down-regulation of the GM-CSF receptor was among the initial events during the specification or commitment processes to CLPs (Kondo et al., 2000). Consistent with this data, when GM-CSF receptor or M-CSF receptor was overexpressed in CLPs, the cells were reprogramed to monocytes or granulocyts by GM-CSF or M-CSF in vitro (Kondo et al., 2000). When a chimeric human IL-3/GM-CSFR chain composed of the extracellular domain of IL-3 and the cytoplasmic domain of GM-CSFR was introduced into murine IL-3-dependent multipotent FDCP-mix cells, the administration of hIL-3 induced cells to granulocytic/moncytic differentiation instead of cell renewal, suggesting that cytoplasmic part of GM-CSFR chain participates in delivering specific signal leading to differentiation (Evans et al., 1999). But this cytokine-induced cell fate convention was not observed in the case where a foreign Epo receptor was overexpressed. Again in another study, when GSF-R transgenic mice were generated under the control of an ubiquitous MHC H-2L promoter, G-CSF was found of capable of stimulating the growth of multiple hematopoietic lineage cells, including BM blast cells from transgenic mice, not only restricted to the neutrophils (Yang et al., 1998).

IL-7, which is produced by BM stromal cells, thymic and intestinal epithelial cells, binds to IL-7 receptor chain and induces its association with _c chain, recruiting Jak1 and Jak2, which in turn activates Pyk2 to prompt lymphoid cell survival (Benbernou et al., 2000). But IL-7 seems to be a unlikely primary factor to induce the commitment of CLPs from HSCs, for in the fetal liver of IL-7^-/- mice, the development of earliest unipotent B cell precursors with phenotype Lin^-CD19^-C-kit⁺IL-7R⁺AA4.1⁺ was found normal, while it was severely affected in SDF-1^-/- embryos (Egawa et al., 2001).

Regulation of commitment of CMPs to MEPs and GMPs

Whereas PU.1 and GATA-1 are co-expressed in CMPs, their mutually exclusive expression coincides with further commitment to either granulocytic/monocytic or megakaryocytic/erythroid differentiation (Figure 3). The expression of GATA-1 decreases in granulocytic/monocytic differentiation while that of PU.1 decreases with differentiation to either megakaryocytic or erythroid direction. Moreover, gene targeting experiments have shown that PU.1 is absolutely required for monocytic and B lymphocytic development while GATA-1 for erythroid and megakaryocytic development.

If hematopoietic lineage commitment is preceded by a state of transcriptional promiscuity, one may ask what is the mechanism that reinforces and stabilizes commitment to specific lineages. Studies suggest that the expression of a transcription factor, above a threshold level, has two reinforcing lineage directing effects: (1) auto-upregulation of its own (Chen et al., 1995; Tsai et al., 1991); and (2) down-regulation of alternative transcription factor pathways. In the clearest system studied to date, PU.1 was found to inhibit the transcriptional activity of GATA-1 upon its target genes, and vice versa. The overexpression of PU.1 inhibited DNA binding of GATA-1 to its consensus element on target gene promoters, and this inhibiting effect could be relived by the over-expression of GATA-1. Detailed studies showed that this PU.1-imposed effect was mediated by direct physical contact of its N-terminal part to the GATA-1 C-terminal zinc finger that involves DNA binding. In a reverse way, possibly through the interaction of the c-finger of GATA-1 with ets domain of PU.1, GATA-1 precludes the recruitment of co-activator c-Jun to the 3/4 region of PU.1 in the context of the promoters of its target genes (Nerlov et al., 2000; Zhang et al., 1999). Additional support for cross-modulation in CMP commitment comes from the findings that ectopic expression of PU.1 in erythroid/megakaryocytic leukemia cell K562 could redirect them to differentiate into granulocytes and monocytes, instead of megakaryocytes upon Ras pathway activation, and overexpression of GATA-1 in granulocytic progenitor 32D cells redirects them to megakaryopoiesis (Matsumura et al., 2000). Taken together, these data suggest that the initial ratio of GATA-1/PU.1 protein at an early, critical time point in the CMP determines the subsequent lineage-restricted fate of the cell (Rekhtman et al., 1999; Zhang et al., 1999, 2000).

Intriguingly, disturbances in the proper kinetic orchestration of critical multivalent lineage-directing transcription factors may result in differentiation arrest along CMPs to MEPs or further procession, thus providing a ground state for developing myeloproliferative disorder or leukemia (Moreau-Gachelin et al., 1988). PU.1 transgenic mice develop erythroleukemia, whose evolution proceeds in two steps: First, a clonal population of erythroblasts emerges from a background of severe anemia; subsequently an Epo-independent subclone emerges, resulting in frank leukemia (Barnache et al., 1998). Experiments with MEL cells indicate that maintenance of PU.1-driven erythroleukemia requires the sustained predominance of PU.1 activity over GATA-1 function, as ectopic GATA-1 cDNA overexpression drives the erythroblasts to terminal erythroid differentiation (Rekhtman et al., 1999).

The execution of lineage-committing function of PU.1 might at least partly be mediated through the extrinsic M-CSF/M-CSFR and GM-CSF/GM-CSFR signaling pathways. Expression of the M-CSFR gene c-fms was absolutely dependent on PU.1 and PU.1^-/- progenitor failed to respond to M-CSF stimulation (DeKoter et al., 1998) and the proper GM-CSFR expression also needed PU.1 (Anderson et al., 1998a). When a murine-M-CSF receptor gene was ectopically expressed in murine mutiple-potent EML cells (Tsai et al., 1994), their differentiation potential to erythroid lineage was compromised and the differentiation to granulocyte/monocyte was favored (Pawlak et al., 2000), suggesting an instructive role of M-CSF/M-CSFR. In this relay, instructive signals through cytokine receptors may be delivered, in part, through homeobox transcription factors. GM-CSF stimulation increases HoxA5 expression in BM cells, and HoxA5 antisense oligonucleotide administration inhibits GM-CFU formation while amplifying the generation of BFU-E (Fuller et al., 1999).

Recent findings suggest that the co-expression of GATA-1 and FOG (friend of GATA-1), instead of GATA-1 alone, destines the precursor cells for erythroid and megakaryocytic development (Tsang et al., 1997) (Figure 3). The differentiation-rescuing experiment demonstrated that the interaction between GATA-1 and FOG was required for either erythroid or megakaryocytic differentiation (Crispino et al., 1999; Tsang et al., 1997). In one series of experiments, C/EBP overexpression induced the eosinophilic differentiation of MEP cell line (HD57, corresponding to CMP ?), and GATA-1 expression was sustained while FOG expression was abrogated. When an exogenous FOG was constitutively expressed, differentiation to eosinophil was inhibited, as indicated by the disappearance of the lineage specific antigen ESO47. It was further found that the inhibiting effect of FOG on the ESO47 promoter activation was mediated by the specific interaction of FOG with the NF motif of GATA-1 in the ESO47 promoter (Querfurth et al., 2000). These data suggest that the differential expression of FOG in eosinophils or megakaryocytic/erythroid cells directly influences the cell fate choice, and that FOG may act as either a co-activator or a co-repressor to GATA-1 in different cellular and promoter contexts.

In addition to these well-documented roles of PU.1, GATA-1 and M-SCFR in the commitment of GMPs and MEPs, other transcription factors and cytokines have also been implicated in these processes. Overexpression of c-Myb, like PU.1, is able to reprogram the K562 cells differentiate to granulocytic/monocytic direction (Matsumura et al., 2000). Egr-1 (early growth response gene-1) was found to promote macrophage production in the expense of erythroid and granulocytic development (Krishnaraju et al., 2001). Conversely, when SCL was introduced into human hematopoietic CD34⁺ cells, the production of erythroid and megakaryocytic colonies were enhanced (Elwood et al., 1998). TGF-1 might be a potent cytokine that biases the commitment of erythroid at the expense of granulocytic and monocytic development (Drexler et al., 1998; Krystal et al., 1994). The secreted protein, WNT, was shown to inhibit the formation of macrophage, while prompting the production of RBC and monocytes from the enriched avian embryonic bone marrow cells or quail mesodermal stem cells QCE6 (Brandon et al., 2000), suggesting a possible role of WNT pathway in favor of erythroid/megakaryocytic differentiation and blockage of late phase differentiation of monocytes to macrophages.

Commitment to granulocytic differentiation

The differential expression of C/EBP appears to influence the lineage choice of bi-potent G/M precursors to either granulocytic or monocytic direction, while PU.1 is needed for both (Anderson et al., 1998b) (Figure 3). CEBP/ was found preferentially expressed in and required for granulocytic lineage, but not for monocytic lineage (Zhang et al., 1997). Over-expression of CEBP/ alone was sufficient to drive the bipotent U937 cells to differentiate into granulocytes, but to hinder their monocytic differentiation induced by PMA (Radomska et al., 1998).

G-CSF itself does not appear to exert instructive role for granulocytic commitment, but rather triggers their maturation once commitment is complete. GSF receptor knock-in mice were produced in which the cytoplasmic domain of GSF receptor was replaced by that of Tpo receptor. Homozygous mutant mice developed normal numbers of morphologically normal neutrophils, although some GSF-dependent functions, such as chemotaxis and mobilization, were impaired (Semerad et al., 1999). Likewise, Epo's role may also be limited to post-commitment events. When the Epo receptor was forcedly expressed in 32D cells, the addition of Epo effectively induced their granulocytic differentiation as G-CSF would do. It was thought of an evidence for the notion that it is not cytokines, but rather the intrinsic properties of cells that determine the differentiation direction (Harris et al., 1998).

The retinoid receptor RARs/RXRs might act as a differentiation-checkpoint switch at the promyelocytic stage of granulopoiesis. The proper concentration of retinoid acid is required for its binding to RAR/RXR heterodimer when the latter is docking on RARE (retinoid acid response element) in regulatory sequences of its target genes. The result is to transform RAR/RXR repressor into an transcriptional activator by releasing co-repressor and recruiting co-activator to itself, which is necessary for the induction of target genes for granulocytic differentiation (Chen et al., 1997; Lin et al., 1999).

Once granulocyte commitment is triggered to terminal differentiation, these same critical transcription factors continue as key players, by down-modulation of cell cycle progression and proliferation. The C/EBP-induced terminal granulocytic differentiation was preceded by an inhibited G1/S transition in cell cycling, as shown in IL-3-dependent cell line 32D cells (Wang et al., 1999), and when 32D cell proliferation was kept on by the overexpresion of c-Myb, the terminal differentiation of these cells induced by G-CSF was blocked (Oh and Reddy, 1998). Likewise, the overexpression of the cDNA encoding an active form Notch1(mNotch1) in 32D cells promoted the terminal granulocyte production through the activation of downstream RBP-J transcription factor, which was also accompanied by an accumulation of cells in G₀/G₁ phase without substantially affecting the apoptosis (Schroeder and Just, 2000b). It is interesting to note that truncated forms of C/EBP were detected in five AML patients and in all the cases acted as dominant negative form to wild type C/EBP (Pabst et al., 2001). It was recently found that loss of JunB expression alone in murine myeloid cells led to a disorder recapitulating the development of CML, arising from an increased proliferation of granulocytic progenitors (Passegue et al., 2001). Jun B has been shown to inhibit fibroblast proliferation through upregulating the transcription activation of CDK inhibitor p16 (Passegue and Wagner, 2000).

The commitment to monocyte differentiation

Several transcription factors have been implicated in the commitment of monocytes from bipotent precursor GMPs. The interferon consensus sequence binding protein (ICSBP/IRF-8), a putative transcriptional partner of PU.1 (Brass et al., 1999), was recently found to be essential for monocytic development while inhibiting the production of neutrophils from GM-CSF-dependent ICSBP^-/- cell line Tot2 and granulocytic progenitor 32D cells (Tamura et al., 2000). The introduction of a foreign ICSBP into Tot2 cells resulted in the upregulation of transcription factor Egr-1 (Early growth response gene-1) and M-CSFR as well as the downregulation of C/EBP and G-CSFR, without obvious effects on the expression of PU.1 and GM-CSFR. Consistent with this interpretation, the overexpression of Egr-1 alone in myeloid-enriched cells has been shown to promote macrophage differentiation and to inhibit granulocyte differentiation (Krishnaraju et al., 2001). By the way, the bZip transcription factor Maf B is upregulated successively from the multipotent progenitor to macrophage. The overexpression of MafB in transformed myeloblasts led to exclusive production of monocyts/macrophages, and such effect was independent of whether a foreign PU.1 was overexpressed (Kelly et al., 2000).

ICSBP may be very crucial for maintaining normal myelopoiesis over and above its role in monocyte commitment. ICSBP expression was found to be greatly decreased in both CML patients and murine CML-like models generated by transgenic expression of Bcr-Abl fusion gene. ICSBP^-/- mice develop granulocytosis with atypical macrophages, resembling a CML-like disease (Holtschke et al., 1996). Conversely, forced expression of ICSBP inhibited Bcr-Abl-induced myeloproliferation both in vitro and in vivo (Hao and Ren, 2000; Schmidt et al., 1998).

The instructive role of M-CSF/MCSFR signaling in monocytic commitment was further supported by a recent study concerning the identification of a novel specific interacting protein FMIP to cytoplasmic domain of M-CSF receptor c-fms, probably as a transient inhibitor of c-fms signaling (Tamura et al., 1999). When FMIP was overexpressed in G/M bipotent FDC-P1Mac11, the M-CSF-induced monocytic differentiation was prevented, instead all cells differentiated into granulocytes.

As with neutrophils, cell cycle arrest might be one prerequisite for the terminal differentiation of monocytic cells. ICSBP-induced macrophage differentiation coincided with cell growth arrest (Tamura et al., 2000). It was shown that HoxA10 overexpression, via the activation of p21, induced cell cycle arrest and monocytic differentiation of U937 cell (Bromleigh and Freedman, 2000). In the presence of IL-3, the overexpression of GATA-2/estrogen chimera in FDCP cell lines induced the cells to monocytic differentiation, which was also accompanied by cell cycle arrest (Heyworth et al., 1999).

As mentioned previously, monocytes/macrophages could derive from a bipotent precursor for either monocyte or B lymphocytes, and the graded expression of PU.1 might govern the choice of cell fate (DeKoter and Singh, 2000). When Lin^- fetal liver cells from PU.1^+/- embryos were cultivated in the presence of IL-7 and stromal cell S17 for 10-14 days, the majority of cells were pro-B, with less than 10% being macrophage. But with constitutive expression of PU.1 cDNA in PU.1^-/- cells, most cells generated in the same culture system turned into macrophages.

The commitment of MEPs to megakaryocytes and erythrocytes

The developments of megakaryocytes and erythrocytes are closely related and share most of transcription factors for lineage development, such as GATA-1, FOG and NF-E2. PHZ-induced hemolytic anemia in mice provokes not only an accelerated erythropoiesis, but also enhanced megakaryocyte production. In this model, late bipotent MEP Ter119⁺/4A5⁺ cells were isolated from both bone marrow and spleen, with individual cells capable of differentiating into either megakaryocytes or erythrocytes within 24-48 h, in the presence of Tpo or Epo, respectively (Vannucchi et al., 2000).

Recent studies concerning GATA-2, protein kinase C- (PKC-) and extracellular signal-regulated kinases (ERKs) induced by Tpo provided some clues for understanding the process of unilineage commitment from MEPs. Overexpression of GATA-2 in human erythroleukemia K562 cells led to cell growth inhibition and phenotype shift from erythroid cells to megakaryocytes (Ikonomi et al., 2000). The expressions of TpoR and GP Ib/IX as well as GP IIb/IIIa were induced while production of total hemoglobin decreased. ERK1 and ERK2 were differentially induced between megakaryocytic and erythroid differentiations from bipotent cell UT-7/GM, by Tpo and Epo respectively. When an exogenous MEK1 (MAP kinase/ERK kinase 1) inhibitor was administrated, the production of hemoglobin-harboring cells was increased (Uchida et al., 2001). Mutations in the TpoR gene have been repeatedly found in the patients with congenital amegkaryocytic thrombocytopenia and TpoR^-/- mice displayed a phenotype of selective thrombocytopenia (Gurney et al., 1994; Tonelli et al., 2000), indicating its selective role in the commitment or maturation of megakaryocyte. It has been shown that the activation of ERKs by Tpo/TpoR coupling could be mediated by multiple pathways, including Jak/Stat and Ras/Raf/MAPK pathways as well as PI3K and PKCzeta (Rojnuckarin et al., 2001). A positive role of PKC in the megakaryocytic commitment has already been established by numerous studies and this effect is at least partly mediated through the activation of Raf-MEK-ERK pathway. A recent study suggested that PKC-, but not other PKC isoforms, functionally cooperated with GATA-1 to activate the expression of megakaryocytic specific antigen IIb (Racke et al., 2001).

Once the erythroid versus megakaryocytic decision is made, numerous transcription factors and cytokine signaling pathways combine to support survival, proliferation, and the expression of lineage specific genes. For example, GATA-1 maintains the expression of Bcl-x and Epo-R, thereby promotes the survival and proliferation of erythroid precursors (Kapur and Zhang, 2001). One of target genes for Epo in this pathway was identified as CHOP, a distant C/EBP family member, involving in the erythroid growth and differentiation (Coutts et al., 1999). At least within the erythroid system, acetylation is one mechanism whereby transcription factors themselves are regulated. CREB-binding protein/p300 binds to GATA-1 and upregulate its DNA binding activity through acetylation (Blobel et al., 1998; Boyes et al., 1998). Similarly, the erythroid inductive effects of SCL, requiring its dissociation from the transcriptional co-repressor mSin3A, was also associated with its acetylated form. In DMSO-induced erythroid differentiation from leukemia cell MEL, SCL was found to be acetylated by P/CAF, which in turn increased SCL's binding affinity with its target. This effect was inhibited by overexpression of mSin3A and cooperated by administration of TSA (Huang and Brandt, 2000; Huang et al., 2000).

The development of T lymphocytes and B lymphocytes from CLPs

It is still unclear whether or not the CLPs with the phenotype of IL-7R⁺Sca-1^lowC-kit^LowLin^- are only immediate precursors for T or B lymphoid cells (Akashi et al., 2000a), or whether macrophages can also be produced from these same cells (Montecino-Rodriguez et al., 2001). As with monocytes, dendritic cells can be derived from either monocyte or lymphoid pathway (Liu et al., 2001).

Environmental signals might play a critical role instructing the commitment of CLPs to either T lymphocytes or B lymphocytes. The development of T lymphocyte needs either CLPs or newly committed early T cells firstly to migrate and home to thymus. The earliest thymic progenitors are CD4^loCD8^-CD44⁺CD25^-C-kit⁺ cells, which are capable of differentiating into B, NK and thymic dendritic cells at a low frequency (Wu et al., 1991), suggesting this population might contain a small number of CLPs.

Recent experiments studying Notch1 suggest one such mechanism to account for T or B cell commitment. Notch1 signaling is not only required for T lymphocyte development, but also promote commitment of CLPs to thymocytes while inhibiting B cell production (Radtke et al., 1999; Pui et al., 1999). Constitutive expression of active Notch1 led to the production of CD4⁺CD8⁺ T cells in bone marrow, suggesting the possibility that activation of Notch 1 is normally induced by thymus but not BM mesenchyme, thereby inducing T cells in thymus and allowing B cell development in BM (Deftos and Bevan, 2000).

T lymphocytes

Notch1 specifies T cell fate by inhibiting B cell lineage specific gene expression, inducing B cells to apoptosis and arresting cell cycling of granulocyte progenitors. It was first found that Notch1 or Notch2 signaling pathway inhibited the transcriptional activity of E2A/E47, an initiating transcription factor for the alternative B cell fate specification, possibly through their inhibiting effect on Ras pathway that is required for the full function of E47 (Ordentlich et al., 1998). More recently it was found that Notch1 signaling not only suppressed B cell IgH gene expression (target gene of E2A), but also induced apoptosis of chicken B cells via the upregulation of Hairy 1, and G1 cell cycle arrest through other pathways (Morimura et al., 2000, 2001). In another study, it was shown that Notch signaling also suppressed the proliferation of granulocytic progenitor 32D cells, by arresting cells in the G0/G1 phase (Schroeder and Just, 2000a), suggesting that Notch1 prevents or terminates 'promiscuous trans-differentiation' of committed T to myeloid cells. On the other hand, it appears that the level of Notch1 expression must be precisely controlled to allow the normal maturation of thymocytes. Constitutive overexpression Notch1 blocks the differentiation of CD4⁺CD8⁺ T cells to CD4⁺ and CD8⁺ T cells late in thymocyte development (Izon et al., 2001).

Ikaros, which in earlier studies was found to play a critical role in the development of all lineage of lymphoid cells, especially NK, T cells and lymphoid dendritic cells (Georgopoulos et al., 1994; Wang et al., 1996), was recently found to set signaling thresholds for pre-TCR- and TCR-dependent T lymphocyte differentiation (Winandy et al., 1999). GATA-3 may likewise regulate early and late stages of T lymphocyte development. The GATA-3^-/- embryonic stem cells failed to contribute to the formation of thymocytes past CD4^- CD8^- stage (Ting et al., 1996), suggesting a role in early T cell development. The GATA-3 binding site has been found in enhancer/promoter regions of multiple T cell specific genes, such as TCR-, - and - sub-unit gene, CD8 and CD4, as well as IFN- and IL-5 (Yamagata et al., 2000). A GATA-3 mutant, KRR-GATA3 that creates local hypoacetylation, led to defective homing and elongated survival of peripheral T cells (Yamagata et al., 2000).

The recent study concerning the commitment of Th1 from Th (T helper) cells provided an illuminating example for how cytokines control differentiation (Mullen et al., 2001; Szabo et al., 2000). The specification of Th1 cell fate was initiated by transcription factor T-bet, independent of IL-12/Stat4 signaling pathway, with mono-allelic expression of the IFN- gene and the induction of IL-12 receptor expression. The IL-12/Stat4 signaling seemed to provide a secondary signal involving the co-activator CBP, which was only required for sustaining the secretion of INF- by the committed Th1 cells. IL-4, in contrast, was found to be able to inhibit the expression of T-bet, thereby skewing the early specification away from Th1.

B lymphocytes

The early phases of B cell development have been well dissected into several distinctive stages by cell surface phenotype and sequential transcription factor expression (Busslinger et al., 2000) (see Figure 4). E2A and EBF are initial transcription factors that specify the B cell fates from the progenitors (Bain et al., 1994; O'Riordan and Grosschedl, 1999). The pro-B cells from EBF^+/- E2A^+/- mice displayed reduced expression of lymphocyte-specific genes, including Pax5, Rag1 and Rag2. Differently, the E2A deficiency promotes early T lymphocyte expansion (Engel et al., 2001). E2A null mutant mice spontaneously developed monoclonal T lymphoma in thymus between 3-6 months, preceded by a decreased percentage of CD44^lowCD25⁺ cells and increased percentage of CD44^high CD25^- DN cell population (Bain et al., 1997). The ectopic expression of E2A induces apoptosis of T lymphoma (Engel et al., 2001). Taken together, these data suggest a role of E2A to limit early T cell expansion and promote differentiation. In another system, the spontaneous conversion of 70Z/3 pre-B cell lymphocytes to cells with a macrophage-like phenotype was associated with the loss of expression of EBF and other B cell specific genes, which could be reversed by the ectopic expression of E2A gene product E12 (Kee and Murre, 1998). Two alternative-splicing bHLH products of E2A gene, E12 and E47, work in an additive way in supporting the B cell development. Surprisingly, the B cell production from E2A deficiency could also be rescued by knocking-in of two- but not one- copy of E2A homologue-HEB gene, while HEB knockout mice showed a basically normal B cell development phenotype (Zhuang et al., 1998).

The transcription factor Pax5/BSAP may play an essential and consolidating role in the committing process of lymphoid precursors to B cell lineage. Firstly, it was found that Pax5^-/- B lymphocytes could develop through the pre-pro B and early pro-B stages and were arrested at the pro-B stage before IgH V-DJ gene rearrangement (Nutt et al., 1997) (see Figure 4). Then unexpectedly, Pax5 homozygous deficient pro-B cells were found of capable of reconstituting T lymphocytes in vivo and even of trans-differentiating into various myeloid cells in vitro (Nutt et al., 1999; Rolink et al., 1999). So it was proposed that the full commitment to B cell fate is set around V-DJ recombination and the phase before this checkpoint could be regarded as a priming or specification phase (Rothenberg, 2000). However, the B cell commitment might happen before pro-B stage (Allman et al., 1999), for Pax5 is expressed in the earliest stage of recognizable B cell population (Nutt and Busslinger, 1999) and its function might not be only restricted to VDJ recombination. For example, Pax5 could consolidate B cell commitment at pro-B by acting as a reverter or terminator to the cells that accidentally slip into monocyte pathway. Additionally, the transcriptional activity of PU.1 on its target genes was compromised by the presence of Pax5 (Maitra and Atchison, 2000), while a higher expression level of PU.1 favors macrophage fate rather than B cell (DeKoter and Singh, 2000), though both cell types require PU.1 for early development. The EML cell line, created by overexpression of a dominant negative form of RAR in primitive BM cells, expresses some marker associated with B cell, and it could be induced to terminal granulocytes by high concentration of ATRA. But when Pax5 was forcedly over-expressed in this cell line, cellular proliferation and myeloid terminal differentiation induced by ATRA was inhibited (Chiang and Monroe, 1999). In support of this, the expression of M-CSFR and MPO in Pax5^-/- cells was efficiently repressed by the introduction of an ectopically expressed Pax5 cDNA (Nutt et al., 1999). Conversely, in the Pax^-/- mice, the B220⁺ cell production size was greatly compromised. Pax5 was found to be transcribed monoallelically at the onset of B cell development, and then transcribed from both alleles at stages of pre-B and immature B cells, and finally was reversed to monoallelic transcription at mature B cells (Nutt and Busslinger, 1999). The monoallelic transcripts of Pax5 mRNA were found to be randomly copied from either one of two parental alleles. Individual cell colonies were able to switch expression between two alleles within 2 weeks and both Pax5 alleles were synchronously replicated during the S phase. This Pax5 expression activation pattern is thought to be compatible with the stochastic model regarding hematopoietic lineage commitment mechanism.

Regulation of HSC differentiation by the cytokine network

Modulation of HSC transcription factor expression by hematopoietic cytokines appears to reflect two overlapping layers of regulation. First, basal hematopoiesis, sufficient to maintain normal basal blood cell production, occurs in the context of low level secretion of multiple cytokines. Overlaid upon this basal combination of cytokine production are short-lived amplifications of specific cytokine secretions, in response to specific hematopoietic stresses (Figure 5).

The cytokine matrix supporting basal hematopoiesis appears to be constructed locally, through the secretion and cell surface presentation of cytokines from mesenchymal cells within the bone marrow. These so-called 'stromal' cells include specialized fibroblasts, endothelial cells, osteoblasts, and perhaps adipocytes. SCF, Tpo, Flt3L and GM-CSF are all produced from within this stromal compartment, primarily from a small, highly proliferative subset of stromal fibroblasts which display cell surface markers of smooth muscle cells as well. Osteoblasts produce both GM-CSF and G-CSF, and may account for the stimulation of basal neutrophil production in vivo. Each of these cytokines is produced at relatively low levels, such that in vitro cultures of these stromal cells only accumulate sub-nanogram/ml concentrates of each cytokine, in which each provides only minimal proliferative stimulus to CD34⁺ cells. Within the context of in vivo bone marrow, these cytokines are undoubtedly locally concentrated, and their concerted activities clearly support HSC survival and proliferation.

SCF, which is expressed constitutively in BM stromal fibroblasts and endothelial cells, is clearly essential to the survival, proliferation, adhesion/migration and differentiation of HSC (Broudy, 1997; Hartman et al., 2001). Absence of SCF protein or its receptor C-kit, as displayed in Sl mutation and W mutation respectively, leads to severe macrocytic anemia. In adult mice, as early as 2 days after administration of a neutralizing antibody against C-kit, all progenitor cells were depleted, and eventually mature myeloid and erythroid cells in BM were absent (Ogawa et al., 1991). Of note, the SCF mRNA is alternatively spliced so that either a soluble or a cell membrane-bound form could be produced. Interestingly, the Sl^d genotype, in which the cytoplasmic part is deleted and so soluble SCF could be normally produced, failed to rescue the macrocytic anemia caused by Sl mutation, suggesting an essential physical contact between stromal cells and hematopoietic progenitors. This observation was further supported by a recent study showing that altered presentation of membrane-bound SCF caused by an amino acid mutation in its cytoplasmic part is associated with a reduced SCF responsiveness by erythroid progenitors (Kapur et al., 1999).

Flt3 L, which is also widely expressed on mesenchymal stromal cells, is the ligand of Flit3R that is restrictedly expressed in CD34⁺ cells but not CD34^- cells within BM or cord blood (Broudy et al., 1996). The targeted disruption of Flt3 causes significant reduction in BM hematopoietic progenitor pool size and subsequently decreased amounts of mature myeloid, B, NK and dendritic cells. Tpo, which was originally found to be synthesized by liver and kidney, has likewise been recently found to be locally produced by bone marrow stromal cells (Sakamaki et al., 1999).

Amplified hematopoiesis in response to physiologic need appears to be regulated in true endocrine fashion, by the stimulated secretion of cytokines disparate from the bone marrow, which then signals the marrow by combined diffusion and bulk plasma transport. For example, anemia or hypoxia triggers increased production of Epo from kidney and liver, resulting an elevated serum Epo level. Since baseline Epo serum concentration only achieves minimal Epo dose-response at the progenitor level, any increase in Epo above this baseline stimulates erythropoiesis. This Epo-dependent erythroid expansion occurs on the background milieu of basal stromal cytokines, and is in fact dependent on SCF/C-kit coupling (Broudy et al., 1996).

Similar induced hematopoietic networks regulate other lineages, and perhaps stem cell self-renewal as well. Acute bacterial infection stimulates monocytes to secrete IL-1 and TNF. These monokines, produced locally at the sites of infection, travel to the bone marrow where they profoundly stimulate the production and secretion of GM-CSF and G-CSF from stromal fibroblasts and endothelium. These cytokines, in turn, stimulate multiple stages of granulopoiesis and also mobilize mature neutrophils into the peripheral blood, thereby causing the paradigmatic clinical manifestation of granulocytosis. In patients with aplastic anemia or cancers receiving chemotherapy, the serum levels of Flt3L fluctuated inversely to the extents of BM failure. Whether Tpo is induced in response to thrombocytopenia is less clear, but it is clearly induced by inflammation, perhaps by TGF-1 and T cell cytokine production. Interestingly, in idiopathic thrombopenic purpura, it is found that the elevated TGF-1 level increased Tpo production from stromal cells, which in turn triggered the expression of TGF receptor on megakaryocytic progenitors, rending them susceptible to the inhibition effect of TGF-1 on differentiation (Sakamaki et al., 1999). The CD40-ligand expressed on activated CD4⁺ T cells enhances the myelopoiesis/megakaryopoiesis, which was found to be mediated by the its stimulating effect on the production of Flt3 by a variety of cell types and of Tpo by stromal cells (Solanilla et al., 2000).

Concluding remarks

Intensive investigations in many genetic and cellular systems over the past two decades have generated a rich view of the process of hematopoietic differentiation. With at least some of the molecular and cellular players now identified, we are now in position to address truly fundamental questions of stem cell biology, including: What is the true cellular potential of primitive stem cells found within the bone marrow, and what directs their commitment to hematopoietic versus non-hematopoietic tissues? What are the earliest events that control HSCs renewal or commitment, and is there any transcription factor specifically maintaining stem cell self-renewal status, or is it maintained simply by cell cycling status? What are environmental signals that critically control the renewal/differentiation and survival of HSCs in BM? Are there specific environmental triggers for stem cell symmetric division, as suggested by, e.g. sperm stem cells? How are kinetic changes in individual transcription factors coordinated? For example, how can PU.1 coexist with GATA-1 in CMPs and direct the restricted choice of either GMPs or EMPs? How do changes in specific transcription factor levels, and changes in local chromatin organization at their loci cause the dysregulated differentiation and genetic instability of myeloproliferative disorders and leukemias? The answers to these questions will establish the foundation for the directed production of multiple hematopoietic and non-hematopoietic cells and tissues, and for the directed treatment of hematopoietic disorders over the decades to come.

Acknowledgements

We thank Diane Giannola for assistance in graphical design.

Akashi K, Reya T, Dalma-Weiszhausz D, Weissman IL. (2000a). Curr. Opin. Immunol., 12: 144-150. Article MEDLINE

Akashi K, Traver D, Miyamoto T, Weissman IL. (2000b). Nature, 404: 193-197. Article MEDLINE

Allman D, Li J, Hardy RR. (1999). J. Exp. Med., 189: 735-740. MEDLINE

Anderson KL, Smith KA, Conners K, McKercher SR, Maki RA, Torbett BE. (1998a). Blood, 91: 3702-3710. MEDLINE

Anderson KL, Smith KA, Pio F, Torbett BE, Maki RA. (1998b). Blood, 92: 1576-1585. MEDLINE

Antonchuk J, Sauvageau G, Humphries RK. (2001). Exp. Hematol., 29: 1125-1134. MEDLINE

Bain G, Engel I, Robanus Maandag EC, te Riele HP, Voland JR, Sharp LL, Chun J, Huey B, Pinkel D, Murre C. (1997). Mol. Cell. Biol., 17: 4782-4791. MEDLINE

Bain G, Maandag EC, Izon DJ, Amsen D, Kruisbeek AM, Weintraub BC, Krop I, Schlissel MS, Feeney AJ, van Roon M et al. (1994). Cell, 79: 885-892. MEDLINE

Barnache S, Wendling F, Lacombe C, Denis N, Titeux M, Vainchenker W, Moreau-Gachelin F. (1998). Oncogene, 16: 2989-2995. MEDLINE

Batard P, Monier MN, Fortunel N, Ducos K, Sansilvestri-Morel P, Phan T, Hatzfeld A, Hatzfeld JA. (2000). J. Cell. Sci., 113: 383-390. MEDLINE

Benbernou N, Muegge K, Durum SK. (2000). J. Biol. Chem., 275: 7060-7065. MEDLINE

Bhardwaj G, Murdoch B, Wu D, Baker DP, Williams KP, Chadwick K, Ling LE, Karanu FN, Bhatia M. (2001). Nat. Immunol., 2: 172-180. Article MEDLINE

Bhatia M, Bonnet D, Wu D, Murdoch B, Wrana J, Gallacher L, Dick JE. (1999). J. Exp. Med., 189: 1139-1148. MEDLINE

Bjornson CR, Rietze RL, Reynolds BA, Magli MC, Vescovi AL. (1999). Science, 283: 534-537. Article MEDLINE

Blazsek I, Chagraoui J, Peault B. (2000). Blood, 96: 3763-3771. MEDLINE

Blobel GA, Nakajima T, Eckner R, Montminy M, Orkin SH. (1998). Proc. Natl. Acad. Sci. USA, 95: 2061-2066. Article MEDLINE

Boyes J, Byfield P, Nakatani Y, Ogryzko V. (1998). Nature, 396: 594-598. Article MEDLINE

Brandon C, Eisenberg LM, Eisenberg CA. (2000). Blood, 96: 4132-4141. MEDLINE

Brass AL, Zhu AQ, Singh H. (1999). EMBO J, 18: 977-991. Article MEDLINE

Brazelton TR, Rossi FM, Keshet GI, Blau HM. (2000). Science, 290: 1775-1779. Article MEDLINE

Bromleigh VC, Freedman LP. (2000). Genes Dev., 14: 2581-2586. Article MEDLINE

Broudy VC. (1997). Blood, 90: 1345-1364. MEDLINE

Broudy VC, Lin NL, Priestley GV, Nocka K, Wolf NS. (1996). Blood, 88: 75-81. MEDLINE

Bryder D, Jacobsen SE. (2000). Blood, 96: 1748-1755. MEDLINE

Busslinger M, Nutt SL, Rolink AG. (2000). Curr. Opin. Immunol., 12: 151-158. Article MEDLINE

Chen H, Ray-Gallet D, Zhang P, Hetherington CJ, Gonzalez DA, Zhang DE, Moreau-Gachelin F, Tenen DG. (1995). Oncogene, 11: 1549-1560. MEDLINE

Chen Z, Wang ZY, Chen SJ. (1997). Pharmacol. Ther., 76: 141-149. MEDLINE

Cheng T, Rodrigues N, Dombkowski D, Stier S, Scadden DT. (2000a). Nat. Med., 6: 1235-1240. Article MEDLINE

Cheng T, Rodrigues N, Shen H, Yang Y, Dombkowski D, Sykes M, Scadden DT. (2000b). Science, 287: 1804-1808. Article MEDLINE

Chiang MY, Monroe JG. (1999). Blood, 94: 3621-3632. MEDLINE

Choi K, Kennedy M, Kazarov A, Papadimitriou JC, Keller G. (1998). Development, 125: 725-732. MEDLINE

Ciau-Uitz A, Walmsley M, Patient R. (2000). Cell, 102: 787-796. MEDLINE

Coutts M, Cui K, Davis KL, Keutzer JC, Sytkowski AJ. (1999). Blood, 93: 3369-3378. MEDLINE

Crispino JD, Lodish MB, MacKay JP, Orkin SH. (1999). Mol. Cell., 3: 219-228. MEDLINE

Cumano A, Godin I. (2001). Curr. Opin. Immunol., 13: 166-171. MEDLINE

Daga A, Podesta M, Capra MC, Piaggio G, Frassoni F, Corte G. (2000). Exp. Hematol., 28: 569-574. MEDLINE

Deftos ML, Bevan MJ. (2000). Curr. Opin. Immunol., 12: 166-172. Article MEDLINE

DeKoter RP, Singh H. (2000). Science, 288: 1439-1441. Article MEDLINE

DeKoter RP, Walsh JC, Singh H. (1998). EMBO J, 17: 4456-4468. Article MEDLINE

Drexler HG, Meyer C, Zaborski M, Uphoff CC, Quentmeier H. (1998). Leuk. Res., 22: 927-938. MEDLINE

Ducos K, Panterne B, Fortunel N, Hatzfeld A, Monier MN, Hatzfeld J. (2000). J. Cell. Physiol., 184: 80-85. Article MEDLINE

Dybedal I, Bryder D, Fossum A, Rusten LS, Jacobsen SE. (2001). Blood, 98: 1782-1791. Article MEDLINE

Egawa T, Kawabata K, Kawamoto H, Amada K, Okamoto R, Fujii N, Kishimoto T, Katsura Y, Nagasawa T. (2001). Immunity, 15: 323-334. MEDLINE

Elwood NJ, Zogos H, Pereira DS, Dick JE, Begley CG. (1998). Blood, 91: 3756-3765. MEDLINE

Engel I, Johns C, Bain G, Rivera RR, Murre C. (2001). J. Exp. Med., 194: 733-746. MEDLINE

Enver T, Greaves M. (1998). Cell, 94: 9-12. MEDLINE

Evans CA, Pierce A, Winter SA, Spooncer E, Heyworth CM, Whetton AD. (1999). Blood, 94: 1504-1514. MEDLINE

Francis JM, Heyworth CM, Spooncer E, Pierce A, Dexter TM, Whetton AD. (2000). J. Biol. Chem., 275: 39137-39145. MEDLINE

Fujmaki S, Harigae H, Sugawara T. (2001). Br. J. Haematol., 113: 52-57. Article MEDLINE

Fuller J, McAdara J, Yaron Y. (1999). Blood, 93: 3391-3400. MEDLINE

Georgopoulos K, Bigby M, Wang J. (1994). Cell, 79: 143-156. MEDLINE

Giampaolo A, Sterpetti P, Bulgarini D. (1994). Blood, 84: 3637-3647. MEDLINE

Giannola D, Shlomchik W, Jegathesan M. (2000). J. Exp. Medicine, 192: 1479-1490.

Gurney AL, Carver-Moore K, de Sauvage FJ, Moore MW. (1994). Science, 265: 1445-1447. MEDLINE

Gussoni E, Soneoka Y, Strickland CD, Buzney EA, Khan MK, Flint AF, Kunkel LM, Mulligan RC. (1999). Nature, 401: 390-394. Article MEDLINE

Hao SX, Ren R. (2000). Mol. Cell. Biol., 20: 1149-1161. MEDLINE

Harris KW, Hu XJ, Schultz S, Arcasoy MO, Forget BG, Clare N. (1998). Blood, 92: 1219-1224. MEDLINE

Hartman M, Piliponsky AM, Temkin V, Levi-Schaffer F. (2001). Blood, 97: 1086-1091. Article MEDLINE

Helgason CD, Sauvageau G, Lawrence HJ, Largman C, Humphries RK. (1996). Blood, 87: 2740-2749. MEDLINE

Heyworth C, Gale K, Dexter M, May G, Enver T. (1999). Genes Dev., 13: 1847-1860. MEDLINE

Holtschke T, Lohler J, Kanno Y, Fehr T, Giese N, Rosenbauer F, Lou J, Knobeloch KP, Gabriele L, Waring JF, Bachmann MF, Zinkernagel RM, Morse IIIHC, Ozato K, Horak I. (1996). Cell, 87: 307-317. MEDLINE

Hsu HC, Ema H, Osawa M, Nakamura Y, Suda T, Nakauchi H. (2000). Blood, 96: 3757-3762. MEDLINE

Hu M, Krause D, Greaves M, Sharkis S, Dexter M, Heyworth C, Enver T. (1997). Genes Dev., 11: 774-785. MEDLINE

Huang S, Brandt SJ. (2000). Mol. Cell. Biol., 20: 2248-2259. MEDLINE

Huang S, Qiu Y, Shi Y, Xu Z, Brandt SJ. (2000). EMBO J., 19: 6792-6803. MEDLINE

Ichikawa Y. (1970). J. Cell. Physiol., 76: 175-184. MEDLINE

Ikonomi P, Rivera CE, Riordan M, Washington G, Schechter AN, Noguchi CT. (2000). Exp. Hematol., 28: 1423-1431. MEDLINE

Izon DJ, Punt JA, Xu L, Karnell FG, Allman D, Myung PS, Boerth NJ, Pui JC, Koretzky GA, Pear WS. (2001). Immunity, 14: 253-264. MEDLINE

Jackson KA, Mi T, Goodell MA. (1999). Proc. Natl. Acad. Sci. USA, 96: 14482-14486. Article MEDLINE

Jakubowiak A, Pouponnot C, Berguido F, Frank R, Mao S, Massague J, Nimer SD. (2000). J. Biol. Chem., 275: 40282-40287. MEDLINE

Just U, Stocking C, Spooncer E, Dexter TM, Ostertag W. (1991). Cell, 64: 1163-1173. MEDLINE

Kapur R, Cooper R, Xiao X, Weiss MJ, Donovan P, Williams DA. (1999). Blood, 94: 1915-1925. MEDLINE

Kapur R, Zhang L. (2001). J. Biol. Chem., 276: 1099-1106. MEDLINE

Karanu FN, Murdoch B, Gallacher L, Wu DM, Koremoto M, Sakano S, Bhatia M. (2000). J. Exp. Med., 192: 1365-1372. MEDLINE

Kee BL, Murre C. (1998). J. Exp. Med., 188: 699-713. MEDLINE

Kelly LM, Englmeier U, Lafon I, Sieweke MH, Graf T. (2000). EMBO J., 19: 1987-1997. Article MEDLINE

Kiger AA, White-Cooper H, Fuller MT. (2000). Nature, 407: 750-754. Article MEDLINE

Klug CA, Morrison SJ, Masek M, Hahm K, Smale ST, Weissman IL. (1998). Proc. Natl. Acad. Sci. USA, 95: 657-662. Article MEDLINE

Koipally J, Renold A, Kim J, Georgopoulos K. (1999). EMBO J., 18: 3090-3100. MEDLINE

Kondo M, Scherer DC, Miyamoto T, King AG, Akashi K, Sugamura K, Weissman IL. (2000). Nature, 407: 383-386. Article MEDLINE

Kondo M, Weissman IL, Akashi K. (1997). Cell, 91: 661-672. MEDLINE

Krause DS, Theise ND, Collector MI, Henegariu O, Hwang S, Gardner R, Neutzel S, Sharkis SJ. (2001). Cell, 105: 369-377. MEDLINE

Krishnaraju K, Hoffman B, Liebermann DA. (2001). Blood, 97: 1298-1305. Article MEDLINE

Krosl J, Baban S, Krosl G, Rozenfeld S, Largman C, Sauvageau G. (1998). Oncogene, 16: 3403-3412. MEDLINE

Krystal G, Lam V, Dragowska W, Takahashi C, Appel J, Gontier A, Jenkins A, Lam H, Quon L, Lansdorp P. (1994). J. Exp. Med., 180: 851-860. MEDLINE

Lacaud G, Carlsson L, Keller G. (1998). Immunity, 9: 827-838. MEDLINE

Lagasse E, Connors H, Al-Dhalimy M, Reitsma M, Dohse M, Osborne L, Wang X, Finegold M, Weissman IL, Grompe M. (2000). Nat. Med., 6: 1229-1234. Article MEDLINE

Lin RJ, Egan DA, Evans RM. (1999). Trends Genet., 15: 179-184. Article MEDLINE

Liu YJ, Kanzler H, Soumelis V, Gilliet M. (2001). Nat. Immunol., 2: 585-589. Article MEDLINE

Maeno M, Mead PE, Kelley C, Xu RH, Kung HF, Suzuki A, Ueno N, Zon LI. (1996). Blood, 88: 1965-1972. MEDLINE

Maguer-Satta V, Oostendorp R, Reid D, Eaves CJ. (2000). Blood, 96: 4118-4123. MEDLINE

Maitra S, Atchison M. (2000). Mol. Cell. Biol., 20: 1911-1922. MEDLINE

Manaia A, Lemarchandel V, Klaine M, Max-Audit I, Romeo P, Dieterlen-Lievre F, Godin I. (2000). Development, 127: 643-653. MEDLINE

Marshall CJ, Kinnon C, Thrasher AJ. (2000). Blood, 96: 1591-1593. MEDLINE

Matsumura I, Kawasaki A, Tanaka H, Sonoyama J, Ezoe S, Minegishi N, Nakajima K, Yamamoto M, Kanakura Y. (2000). Blood, 96: 2440-2450. MEDLINE

Medvinsky A, Dzierzak E. (1996). Cell, 86: 897-906. MEDLINE

Meng X, Lindahl M, Hyvonen ME, Parvinen M, de Rooij DG, Hess MW, Raatikainen-Ahokas A, Sainio K, Rauvala H, Lakso M, Pichel JG, Westphal H, Saarma M, Sariola H. (2000). Science, 287: 1489-1493. Article MEDLINE

Mezey E, Chandross KJ, Harta G, Maki RA, McKercher SR. (2000). Science, 290: 1779-1782. Article MEDLINE

Montecino-Rodriguez E, Leathers H, Dorshkind K. (2001). Nat. Immunol., 2: 83-88. Article MEDLINE

Moreau-Gachelin F, Tavitian A, Tambourin P. (1988). Nature, 331: 277-280. MEDLINE

Morimura T, Goitsuka R, Zhang Y, Saito I, Reth M, Kitamura D. (2000). J. Biol. Chem., 275: 36523-36531. MEDLINE

Morimura T, Miyatani S, Kitamura D, Goitsuka R. (2001). J. Immunol., 166: 3277-3283. MEDLINE

Mullen A, High F, Hutchins A. (2001). Science, 292: 1907-1910. Article MEDLINE

Nerlov C, Graf T. (1998). Genes Dev., 12: 2403-2412. MEDLINE

Nerlov C, Querfurth E, Kulessa H, Graf T. (2000). Blood, 95: 2543-2551. MEDLINE

Nichogiannopoulou A, Trevisan M, Neben S, Friedrich C, Georgopoulos K. (1999). J. Exp. Med., 190: 1201-1214. MEDLINE

Niwa H, Burdon T, Chambers I, Smith A. (1998). Genes Dev., 12: 2048-2060. MEDLINE

Nutt SL, Busslinger M. (1999). Biol. Chem., 380: 601-611. MEDLINE

Nutt SL, Heavey B, Rolink AG, Busslinger M. (1999). Nature, 401: 556-562. Article MEDLINE

Nutt SL, Urbanek P, Rolink A, Busslinger M. (1997). Genes Dev., 11: 476-491. MEDLINE

O'Riordan M, Grosschedl R. (1999). Immunity, 11: 21-31. MEDLINE

Ogawa M, Matsuzaki Y, Nishikawa S, Hayashi S, Kunisada T, Sudo T, Kina T, Nakauchi H. (1991). J. Exp. Med., 174: 63-71. MEDLINE

Oh IH, Reddy EP. (1998). Mol. Cell. Biol., 18: 499-511. MEDLINE

Ohneda O, Fennie C, Zheng Z, Donahue C, La H, Villacorta R, Cairns B, Lasky LA. (1998). Blood, 92: 908-919. MEDLINE

Okuda T, Takeda K, Fujita Y, Nishimura M, Yagyu S, Yoshida M, Akira S, Downing JR, Abe T. (2000). Mol. Cell. Biol., 20: 319-328. MEDLINE

Okuda T, van Deursen J, Hiebert SW, Grosveld G, Downing JR. (1996). Cell, 84: 321-330. MEDLINE

Ordentlich P, Lin A, Shen CP, Blaumueller C, Matsuno K, Artavanis-Tsakonas S, Kadesch T. (1998). Mol. Cell. Biol., 18: 2230-2239. MEDLINE

Osawa M, Hanada K, Hamada H, Nakauchi H. (1996). Science, 273: 242-245. MEDLINE

Pabst T, Mueller BU, Zhang P, Radomska HS, Narravula S, Schnittger S, Behre G, Hiddemann W, Tenen DG. (2001). Nat. Genet., 27: 263-270. Article MEDLINE

Pan Q, Simpson RU. (1999). J. Biol. Chem., 274: 8437-8444. MEDLINE

Pan Q, Simpson RU. (2001). J. Endocrinol., 169: 153-159. MEDLINE

Papayannopoulou T, Priestley GV, Rohde A, Peterson KR, Nakamoto B. (2000). Blood, 95: 1274-1282. MEDLINE

Passegue E, Jochum W, Schorpp-Kistner M, Mohle-Steinlein U, Wagner EF. (2001). Cell, 104: 21-32. MEDLINE

Passegue E, Wagner EF. (2000). EMBO J., 19: 2969-2979. MEDLINE

Pawlak G, Grasset MF, Arnaud S, Blanchet JP, Mouchiroud G. (2000). Exp. Hematol., 28: 1164-1173. MEDLINE

Peschon JJ, Morrissey PJ, Grabstein KH, Ramsdell FJ, Maraskovsky E, Gliniak BC, Park LS, Ziegler SF, Williams DE, Ware CB et al. (1994). J. Exp. Med., 180: 1955-1960. MEDLINE

Philips R, Ernst R, Brunk B. (2000). Science, 288: 1635-1640. Article MEDLINE

Pierelli L, Marone M, Bonanno G, Mozzetti S, Rutella S, Morosetti R, Rumi C, Mancuso S, Leone G, Scambia G. (2000). Blood, 95: 3001-3009. MEDLINE

Porcher C, Liao EC, Fujiwara Y, Zon LI, Orkin SH. (1999). Development, 126: 4603-4615. MEDLINE

Pui JC, Allman D, Xu L, DeRocco S, Karnell FG, Bakkour S, Lee JY, Kadesch T, Hardy RR, Aster JC, Pear WS. (1999). Immunity, 11: 299-308. MEDLINE

Querfurth E, Schuster M, Kulessa H, Crispino JD, Doderlein G, Orkin SH, Graf T, Nerlov C. (2000). Genes Dev., 14: 2515-2525. MEDLINE

Racke FK, Wang D, Zaidi Z, Kelley J, Visvader J, Soh JW, Goldfarb AN. (2001). J. Biol. Chem., 276: 522-528. MEDLINE

Radomska HS, Huettner CS, Zhang P, Cheng T, Scadden DT, Tenen DG. (1998). Mol. Cell. Biol., 18: 4301-4314. MEDLINE

Radtke F, Wilson A, Stark G, Bauer M, van Meerwijk J, MacDonald HR, Aguet M. (1999). Immunity, 10: 547-558. MEDLINE

Raz R, Lee CK, Cannizzaro LA, d'Eustachio P, Levy DE. (1999). Proc. Natl. Acad. Sci. USA, 96: 2846-2851. MEDLINE

Rebel VI, Hartnett S, Hill GR, Lazo-Kallanian SB, Ferrara JL, Sieff CA. (1999). J. Exp. Med., 190: 1493-1504. MEDLINE

Rekhtman N, Radparvar F, Evans T, Skoultchi AI. (1999). Genes Dev., 13: 1398-1411. MEDLINE

Robb L, Lyons I, Li R, Hartley L, Kontgen F, Harvey RP, Metcalf D, Begley CG. (1995). Proc. Natl. Acad. Sci. USA, 92: 7075-7079. MEDLINE

Robertson SM, Kennedy M, Shannon JM, Keller G. (2000). Development, 127: 2447-2459. MEDLINE

Rojnuckarin P, Miyakawa Y, Fox NE, Deou J, Daum G, Kaushansky K. (2001). J. Biol. Chem., 4: 4.

Rolink AG, Nutt SL, Melchers F, Busslinger M. (1999). Nature, 401: 603-606. Article MEDLINE

Rothenberg EV. (2000). Curr. Opin. Genet. Dev., 10: 370-379. Article MEDLINE

Sabbattini P, Lundgren M, Georgiou A, Chow C, Warnes G, Dillon N. (2001). EMBO J., 20: 2812-2822. Article MEDLINE

Sakamaki S, Hirayama Y, Matsunaga T, Kuroda H, Kusakabe T, Akiyama T, Konuma Y, Sasaki K, Tsuji N, Okamoto T, Kobune M, Kogawa K, Kato J, Takimoto R, Koyama R, Niitsu Y. (1999). Blood, 94: 1961-1970. MEDLINE

Sauvageau G, Thorsteinsdottir U, Eaves CJ, Lawrence HJ, Largman C, Lansdorp PM, Humphries RK. (1995). Genes Dev., 9: 1753-1765. MEDLINE

Schmidt M, Nagel S, Proba J, Thiede C, Ritter M, Waring JF, Rosenbauer F, Huhn D, Wittig B, Horak I, Neubauer A. (1998). Blood, 91: 22-29. MEDLINE

Schroeder T, Just U. (2000a). Exp. Hematol., 28: 1206-1213.

Schroeder T, Just U. (2000b). EMBO J., 19: 2558-2568. Article MEDLINE

Scott EW, Simon MC, Anastasi J, Singh H. (1994). Science, 265: 1573-1577. MEDLINE

Semerad CL, Poursine-Laurent J, Liu F, Link DC. (1999). Immunity, 11: 153-161. MEDLINE

Shi Q, Rafii S, Wu MH, Wijelath ES, Yu C, Ishida A, Fujita Y, Kothari S, Mohle R, Sauvage LR, Moore MA, Storb RF, Hammond WP. (1998). Blood, 92: 362-367. MEDLINE

Solanilla A, Dechanet J, El Andaloussi A, Dupouy M, Godard F, Chabrol J, Charbord P, Reiffers J, Nurden AT, Weksler B, Moreau JF, Ripoche J. (2000). Blood, 95: 3758-3764. MEDLINE

Szabo SJ, Kim ST, Costa GL, Zhang X, Fathman CG, Glimcher LH. (2000). Cell, 100: 655-669. MEDLINE

Tamura T, Mancini A, Joos H, Koch A, Hakim C, Dumanski J, Weidner KM, Niemann H. (1999). Oncogene, 18: 6488-6495. MEDLINE

Tamura T, Nagamura-Inoue T, Shmeltzer Z, Kuwata T, Ozato K. (2000). Immunity, 13: 155-165. MEDLINE

Thorsteinsdottir U, Sauvageau G, Humphries RK. (1999). Blood, 94: 2605-2612. MEDLINE

Ting CN, Olson MC, Barton KP, Leiden JM. (1996). Nature, 384: 474-478. MEDLINE

Tonelli R, Scardovi AL, Pession A, Strippoli P, Bonsi L, Vitale L, Prete A, Locatelli F, Bagnara GP, Paolucci G. (2000). Hum. Genet., 107: 225-233. MEDLINE

Trinh LA, Ferrini R, Cobb BS, Weinmann AS, Hahm K, Ernst P, Garraway IP, Merkenschlager M, Smale ST. (2001). Genes Dev., 15: 1817-1832. MEDLINE

Tsai FY, Keller G, Kuo FC, Weiss M, Chen J, Rosenblatt M, Alt FW, Orkin SH. (1994). Nature, 371: 221-226. MEDLINE

Tsai SF, Strauss E, Orkin SH. (1991). Genes Dev., 5: 919-931. MEDLINE

Tsang AP, Visvader JE, Turner CA, Fujiwara Y, Yu C, Weiss MJ, Crossley M, Orkin SH. (1997). Cell, 90: 109-119. MEDLINE

Tsuzuki S, Towatari M, Saito H, Enver T. (2000). Mol. Cell. Biol., 20: 6276-6286. MEDLINE

Uchida M, Kirito K, Shimizu R, Miura Y, Ozawa K, Komatsu N. (2001). Int. J. Hematol., 73: 78-83. MEDLINE

Vannucchi AM, Paoletti F, Linari S, Cellai C, Caporale R, Ferrini PR, Sanchez M, Migliaccio G, Migliaccio AR. (2000). Blood, 95: 2559-2568. MEDLINE

Varnum-Finney B, Xu L, Brashem-Stein C, Nourigat C, Flowers D, Bakkour S, Pear WS, Bernstein ID. (2000). Nat. Med., 6: 1278-1281. Article MEDLINE

Wang JH, Nichogiannopoulou A, Wu L, Sun L, Sharpe AH, Bigby M, Georgopoulos K. (1996). Immunity, 5: 537-549. MEDLINE

Wang X, Scott E, Sawyers CL, Friedman AD. (1999). Blood, 94: 560-571. MEDLINE

Winandy S, Wu L, Wang JH, Georgopoulos K. (1999). J. Exp. Med., 190: 1039-1048. MEDLINE

Wu L, Antica M, Johnson GR, Scollay R, Shortman K. (1991). J. Exp. Med., 174: 1617-1627. MEDLINE

Yamada Y, Warren AJ, Dobson C, Forster A, Pannell R, Rabbitts TH. (1998). Proc. Natl. Acad. Sci. USA, 95: 3890-3895. MEDLINE

Yamagata T, Mitani K, Oda H, Suzuki T, Honda H, Asai T, Maki K, Nakamoto T, Hirai H. (2000). EMBO J., 19: 4676-4687. MEDLINE

Yang FC, Watanabe S, Tsuji K, Xu MJ, Kaneko A, Ebihara Y, Nakahata T. (1998). Blood, 92: 4632-4640. MEDLINE

Zhang DE, Zhang P, Wang ND, Hetherington CJ, Darlington GJ, Tenen DG. (1997). Proc. Natl. Acad. Sci. USA, 94: 569-574. Article MEDLINE

Zhang P, Behre G, Pan J, Iwama A, Wara-Aswapati N, Radomska HS, Auron PE, Tenen DG, Sun Z. (1999). Proc. Natl. Acad. Sci. USA, 96: 8705-8710. Article MEDLINE

Zhang P, Zhang X, Iwama A, Yu C, Smith KA, Mueller BU, Narravula S, Torbett BE, Orkin SH, Tenen DG. (2000). Blood, 96: 2641-2648. MEDLINE

Zhao Y, Lin Y, Zhan Y, Yang G, Louie J, Harrison DE, Anderson WF. (2000). Blood, 96: 3016-3022. MEDLINE

Zhou Y, Lim KC, Onodera K, Takahashi S, Ohta J, Minegishi N, Tsai FY, Orkin SH, Yamamoto M, Engel JD. (1998). EMBO J., 17: 6689-6700. MEDLINE

Zhuang Y, Barndt RJ, Pan L, Kelley R, Dai M. (1998). Mol. Cell. Biol., 18: 3340-3349. MEDLINE

Ziegler BL, Valtieri M, Porada GA, De Maria R, Muller R, Masella B, Gabbianelli M, Casella I, Pelosi E, Bock T, Zanjani ED, Peschle C. (1999). Science, 285: 1553-1558. Article MEDLINE

Figure 1 Transcriptional regulation of early hematopoietic stem cell development. HSCs, as one progeny of hemangioblasts or hemogenic endothelium, are faced with the cell fate choice either to self-renew or to differentiate into committed common lymphoid or common myeloid hematopoietic precursors. The transcription factors involved in each development direction are depicted

Figure 2 Extrinsic stem cell hematopoietic growth factors direct HSC transcription programs. For example, the Notch1 ligand Jagged-1 is present on BM mesenchymal cells. Upon binding to membrane-bound receptor Notch1, the cytoplasmic domain of Notch1 is cleaved, allowing it to diffuse to the nucleus where it acts as a transcriptional activator of several HSC gene programs. As another example, BMP-4, which has been implicated to delivering inductive signals for the origin of HSCs in embryo, might act through its effect on the transcriptional regulation of GATA-2

Figure 3 Transcriptional regulation of common myeloid precursor (CMP) commitment. CMPs differentiate into either common precursors for granulocytic and monocytic lineages (GMPs) or common precursors for both erythroid and megakaryocytic lineages (EMPs). A separate, possible, pathway leading to eosinophils is depicted by dotted line. Dual expression of PU.1 and GATA-1 leads HSCs to CMPs, but then dominant expression of PU.1 is restricted to GMPs, while unopposed GATA-1 expression directs differentiation to EMPs

Figure 4 Transcription regulation of common lymphoid precursor (CLP) commitment. B lymphocytes and T lymphocytes are derived from a common lymphoid precursor (CLP). The early development of B cell is distinguished into distinct stages by the sequential expression of different transcription factors that direct Ig gene recombination and the expression of B cell specific cell surface phenotypes. A proposed (alternative) differentiation pathway of macrophages from pro-B is also indicated by a dotted line

Figure 5 Cytokine production in basal and induced (amplified) hematopoiesis. (Top) The survival and slow proliferation of hematopoietic stem cells (HSC) is maintained by the local membrane presentation and secretion of low levels of stem cell factor (SCF), Fl3 ligand (Flt-3l) and thrombopoietin, and likely additional cofactors, produced by mesenchymal cells lining the bone marrow cavity, including fibroblasts (FB), endothelial cells (endo) and osteoblasts (OB). Basal differentiation of HSC to mature granulocytes and monocytes is maintained by the local matrix presentation and secretion of G-CSF and GM-CSF from osteoblasts, Tpo, and the endocrine delivery of low levels of Epo (produced by peritubular endothelial cells in the kidney, and the liver). (Bottom) Lineage-specific amplification of hematopoiesis is stimulated by increased secretion and delivery of lineage-specific cytokines. Increased Epo is produced in the kidney in response to hypoxia or anemia (local hyoxemia), Fb and endo secrete increased quantities of G-CSF and GM-CSF in response to IL-1 and TNF produced at sites of inflammation and infection by activated monocyte/macrophages, and increased Tpo may be produced by Fb in response to TGF, produced by activated T cells or activated Fb themselves