The homeostatic production of circulating blood cells is one of the most dynamic processes in the human body. Nearly 200 billion red blood cells, 10 billion white blood cells, and 400 billion platelets are produced daily throughout our lifetime. In addition to the requirement for high cell production, the concentration of individual blood cell lineages must be maintained within narrow limits in the peripheral blood and tissues. Superimposed on this dynamic system is the requirement of the blood-forming tissues to produce particular cell types in response to specific demands. For example, microbial breach of the skin and mucosal barriers must be met with an increase in the number and antimicrobial properties of granulocytes to protect the host, acute blood loss requires an increase in oxygen-carrying erythroid cells to sustain normal host organ function, and so on. When the tightly regulated production of blood cells fails, the host may develop severe disease as a result of too few (marrow aplasia and anemia) or too many (leukemia) blood cells.
Hematopoiesis is the process of blood cell production. The stem cell theory of hematopoiesis purports that all hematopoietic elements are ultimately derived from a transplantable multilineage long-term repopulating hematopoietic stem cell. This chapter will review some of the experimental evidence, primarily documented in the murine system, that supports this theory and provide an overview of the in vitro and in vivo assays that discriminate self-renewing hematopoietic stem cells from multipotent and lineage committed progenitor and mature cells. The intent of this overview is to provide a basic understanding of the hematopoietic hierarchy. Hopefully, the experimentally defined paradigm for the hierarchy of hematopoietic stem and progenitor cell compartments will prove useful in understanding the biology of stem cells in other organs, including the cardiovascular system.
ORIGINS OF BLOOD CELL INVESTIGATION
Circulating erythrocytes were the first blood cells identified with the aid of the microscope in the late 1600s. The colorless white blood cells, however, remained obscured from direct observation until the late 1800s. With further improvements in microscopes and the use of a variety of cellular stains, a new era in hematology was born at the turn of the 20th century where changes in blood cell number, morphology, or location were now correlated with specific human diseases. Further progress in studying hematopoiesis was largely limited to conjecture and opinion until the mid-20th century when hematologists began to examine the hematopoietic consequences of animal exposure to ionizing radiation (Wintrobe, 1980).
METHODS OF IDENTIFYING HEMATOPOIETIC STEM AND PROGENITOR CELLS
Whole-body irradiation of animals was noted to depress hematopoiesis and cause life-threatening deficiencies in all circulating blood cells (pancytopenia) in exposed mice (Jacobson et al., 1949). Shielding the spleen of the experimental animal from the radiation beam resulted in recovery from an otherwise lethal dose of x-ray exposure (Jacobson et al., 1949). The splenic protection of the irradiated subject was thought to be conferred by a humoral factor since removal of the spleen 1–6 hr after the lethal dose of irradiation was demonstrated to be sufficient for the protection (Jacobson et al., 1951a). However, transplantation of intact spleens into the abdomen of the lethally irradiated hosts also provided radioprotection (Jacobson et al., 1951b). In 1956, Ford et al. (1956) provided definitive evidence that hematopoietic cells and not plasma or subcellular molecules conferred radiation protection. Using donor cells from a strain of mice carrying a balanced (but morphologically identifiable) chromosome translocation, blood cells of a lethally irradiated mouse were demonstrated to be effectively replaced by donor hematopoietic cells after transplantation. Soon after, Till and McCulloch (1961) provided evidence that single multipotent hematopoietic progenitor cells (cells giving rise to more than one lineage of blood cells) could be identified in vivo by injecting donor marrow cells into a lethally irradiated recipient animal and examining the recipient spleen for hematopoietic colonies 8–12 days later. Each colony of hematopoietic cells in the spleen was demonstrated to arise from a single precursor cell, the colony-forming unit in spleen cell (CFU-S) (Becker et al., 1963). Use of this assay provided the first compelling evidence that hematopoietic cells were clonally derived. While this method advanced the study of hematopoietic stem cell biology, little was known in the early 1960s of the mechanisms that caused and/or permitted stem cells to differentiate into the mature cells.
It was clear to experimental hematologists in the mid-1960s that new methods would be required to isolate the clonal precursors of the mature blood cells and to determine the factors that permitted their proliferation, survival, and differentiation. Pluznik and Sachs (1965) and Bradley and Metcalf (1966) reported that murine hematopoietic cells could be cultured in vitro and that addition of soluble fluid from several murine organs (i.e., urine or pregnant uterine extract) resulted in the in vitro formation of myeloid colonies. This biologically active tissue fluid was called colony-stimulating activity (CSA). Furthermore, these investigators reported that each myeloid colony developing in vitro arose from a single precursor cell called the colony-forming unit in culture (CFU-C) (Fig. 1). Further work revealed that myeloid colonies were comprised of granulocytes (CFU-G), macrophages (CFU-M), or both (CFU-GM) and that the type of colony formed was related in part to the type of CSA added. Later, using anemic mouse serum as a source of red blood cell CSA, Axelrad et al. (1974) succeeded in demonstrating that red blood cell colonies were clonally derived in vitro and called the most primitive red cell precursors the erythroid burst-forming unit (BFU-E) and the most committed erythroid CFC were called colony-forming unit erythroid cells (CFU-E). Multipotent progenitor cells were also identified and called CFU-Mix or CFU-GEMM to reflect the colony composition of granulocytes, erythroid cells, mast cells, megakaryocytes, and/or macrophages (Fig. 1).
Plating of hematopoietic cells in special double-layer agar cultures with multiple recombinant cytokines permits the identification of hematopoietic progenitors that are highly proliferative (highly proliferative potential colony-forming cells (HPP-CFC)) (Kriegler et al., 1994). HPP-CFC colonies contain >50,000 cells that are visible in the culture dishes without need for magnification. HPP-CFC can be plucked from the agar medium, dispersed into a single-cell suspension, and replated in standard CFC assays with emergence of CFU-Mix, CFU-GM, and some BFU-E (McNiece et al., 1990). Thus, the HPP-CFC is considered to be a more primitive cell than the multipotent hematopoietic progenitor (CFU-GEMM). In fact, HPP-CFC are the most primitive hematopoietic progenitor cells that can be cultured in vitro without the presence of hematopoietic stromal cells in co-culture. Nonetheless, HPP-CFC are heterogenous progenitors. Several investigators have fully characterized the murine HPP-CFC population and have allocated HPP-CFC into differentiation classes (Kriegler et al., 1994). The most primitive HPP-CFC require four or more growth factors to form colonies, while more committed HPP-CFC will emerge in culture in the presence of only three or as few as two growth factors. The most committed HPP-CFC display a more limited replating potential and are somewhat smaller in diameter with fewer cells comprising the colony.
In the late 1970s, Allen and Dexter (1984) reported on techniques to culture murine bone marrow for prolonged periods in vitro. These cultures were dependent upon the development of complex monolayers of nonhematopoietic stromal cells adherent to the culture plates. Further analysis revealed that the stromal elements were composed of endothelial cells, fibroblasts, macrophages, mesenchymal cells, and adipocytes (Penn et al., 1993). In time, procedures arose that permitted the establishment of primary stromal cell layers that were subsequently treated with DNA cross-linking agents to prevent further cell division but allow cell survival. Isolated populations of hematopoietic stem and progenitor cells could then be added to the mitotically quiescent stromal layers and analysis of hematopoietic cell proliferation and differentiation performed. Such study revealed that the most primitive hematopoietic cells (those giving rise to hematopoietic progenitors for up to 35 days and longer) resided beneath the stromal monolayers and were phase contrast dark (colony area-forming cells (CAFC)), whereas the mature blood cells produced via progenitor cell differentiation were released into the tissue culture medium as nonadherent cells and were phase contrast bright (Ploemacher et al., 1989). Using these morphologic characteristics some investigators determined that the number of day 35 CAFC was found to correlate with the number of long-term repopulating hematopoietic stem cells in vivo (Ploemacher et al., 1989). A modification of this assay called for the co-culture of hematopoietic cells with mitotically inactive primary stromal cells (or some stromal cell lines) and, after varying times in culture, isolation of the cells in co-culture and plating of the cells in standard CFC assays. Those hematopoietic progenitors co-cultured for prolonged periods with retention of the ability to give rise to CFC were called long-term culture-initiating cells (LTC-IC) (Sutherland et al., 1989). Using limiting dilution analysis, the number of such murine or human LTC-IC that give rise to multipotent and committed progenitors for 3–4 weeks in vitro can be calculated (Ponchio et al., 1995; Verfaillie and Miller, 1995; Leemhuis et al., 1996) and are considered to be the most primitive hematopoietic precursor detectable via in vitro assays.
The above CFC assays identify progenitor cells of the myeloerythroid series but not B or T lymphocytes. B lymphocytes require the presence of bone marrow or fetal liver stromal cells to promote proliferation and maturation in vitro (Muller-Sieburg et al., 1986; Whitlock and Witte, 1987). Addition of certain growth factors to the stromal cell cultures improves B cell maturation in vitro (Pietrangeli et al., 1988). T cell clones can be identified in the thymus in irradiated recipient mice following direct injection (Ezine et al., 1984). T lymphocyte progenitor cells can also be identified using fetal thymic organ cultures (Eren et al., 1987). Combining these lymphocyte progenitor assays with the myeloerythroid CFC assays permits identification of essentially all clonal hematopoietic progenitors.
The frequency of human and murine hematopoietic stem cells has been estimated to be 1/104–105 bone marrow cells (Harrison et al., 1989; Sutherland et al., 1989; Harrison, 1993; Szilvassy and Hoffman, 1995; Bhatia et al., 1998; Abkowitz et al., 2002). The rarity of this population presented considerable impediments to isolation until recently. Several significant advances that have led to hematopoietic stem cell identification include development of bone marrow transplantation techniques, production of monoclonal antibodies to stem cell surface antigens, and availability of high-speed cell-sorting techniques to enrich for the rare stem cells. Murine bone marrow transplantation experiments have provided the necessary evidence to reliably identify the murine hematopoietic stem cell. In fact, murine hematopoietic stem cells are now defined as cells that self-renew in vivo and also proliferate and differentiate into all lineages of circulating peripheral blood cells for more than 4 months after transplantation into recipient animals (Orlic and Bodine, 1994).
Murine hematopoietic stem and progenitor cells can be enriched (negative selection) using flow cytometric cell sorting (Table 1) by selecting bone marrow cells that fail to express cell surface antigens (lin–) typically displayed by mature B and T lymphocytes, neutrophils, macrophages, natural killer cells, and red blood cells (Muller-Sieburg et al., 1986). Three commonly used phenotypic markers expressed by stem cells (positive selection) include stem cell antigen-1 (Sca-1), c-kit (CD117), and Thy-1 (CD90) (Morrison et al., 1995). Sca-1 is a cell surface molecule that is required for normal stem cell self-renewal and progenitor cell proliferation and lineage maturation (Ito et al., 2003). C-kit is a cell surface receptor tyrosine kinase that is necessary for hematopoietic stem cell survival and in utero embryo survival (Reith and Bernstein, 1991). Thy-1 is a cell surface molecule expressed by stem and lymphoid cells (Basch and Berman, 1982). CD34 is a cell surface sialomucin expressed by hematopoietic and endothelial cells (Krause et al., 1996). Stem cells in the murine fetal liver highly express CD34 and the C1q complement receptor (AA4.1) (Tavassoli, 1994). These antigens (CD34 and AA4.1) are expressed on proliferating bone marrow stem cells but are downregulated and not detectable on the surface of quiescent marrow stem cells (Szilvassy and Cory, 1993; Sato et al., 2000). Other antigens that have been utilized to enrich (positive or negative) murine bone marrow stem cells include AC133, CD31, CD38, CD43, CD105, and leukocyte function antigen-1 (CD11a). Stem cells may also be enriched from other marrow cells by the fact that quiescent stem cells bind the least amount of a DNA-binding dye (Hoechst 33342) and appear as a side population distinguishable (by cell sorting) from other hematopoietic cells. Retention of low amounts of another vital dye, Rhodamine 123, is also a feature of hematopoietic stem cells. Use of several different combinations of these cell surface markers has permitted enrichment of stem cells to homogeneity as evidenced by long-term repopulation of all blood lineages in recipient mice from a single transplanted cell with a phenotype, Sca-1+c-Kit+CD34–lin–.
Cell surface phenotypic markers on adult murine bone marrow stem cells
Cells expressing the “positive” markers are enriched for hematopoietic stem cell repopulating ability while those expressing the “negative” markers are devoid of stem cell activity. Stem cell isolation protocols generally employ both positive and negative selection strategies to obtain highly enriched cells. Abbreviations: Stem cell antigen-1 (Sca-1), multidrug resistence gene (MDR), granulocyte-1 (Gr-1), and terminal erythroid 119 antigen (TER119). A more detailed description of the monoclonal antibodies used and stem cell selection strategies has recently been published (Srour and Yoder, 2003).
The most potent stem cells display low levels of these antigens and limited nuclear or cytoplasmic accumulation of these vital stains (Hoechst 33342 and Rhodamine 123, respectively).
The hematopoietic system is comprised of a series of functionally and phenotypically distinguishable progenitor cell compartments (Fig. 1). Bone marrow and fetal liver cells isolated using the above strategies can be further fractionated into several distinct progenitor cell populations. C-kit+, Thy-1lo, Sca-1+lin– cells possess long-term repopulating ability, but cells that are c-kit+Thy-1loSca-1+ and Mac-1lo (a granulocyte adhesion molecule) possess only short-term repopulating ability (Morrison and Weissman, 1994). While the long-term repopulating cells can differentiate into the short-term repopulating cells, the opposite is not true (Fig. 2) (Morrison et al., 1997). Downstream of the short-term repopulating progenitors are the common lymphoid (CLP) and common myeloid progenitors (CMP) (Fig. 2) (Kondo et al., 1997; Akashi et al., 2000). These progenitors differ in that the CLP express the interleukin-7 receptor (IL-7R) but not the granulocyte-macrophage colony-stimulating factor receptor (GM-CSFR) while the CMP are GM-CSFR+IL-7R–. The CMP can be further differentiated into committed granulocyte-macrophage progenitors (GMP) or megakaryocyte-erythroid progenitors (MEP) (Fig. 2). GMP express the macrophage colony-stimulating factor receptor (M-CSFR) but not the erythropoietin receptor (Epo-R), while the MEP are M-CSFR–Epo-R+ (Akashi et al., 2000). Recent data demonstrate that lethally irradiated mice can be rescued from radiation-induced pancytopenia by infusion of MEP alone (Nakorn et al., 2002). While the recipient animals are not long-term repopulated by the MEP, these committed progenitors differentiate into platelets and red blood cells that support the animals until surviving (radiation-resistant) endogenous hematopoietic stem cells repopulate all of the peripheral blood cell lineages.
Human hematopoietic cells have also been isolated using monoclonal antibodies and flow cytometric cell sorting (reviewed in Payne and Crooks (2002) and Verfaillie (2002)). Human stem cells have been isolated from bone marrow, umbilical cord blood, fetal liver, and mobilized peripheral blood as cells expressing CD34, Thy-1, AC133, and c-kit but not CD38 or mature blood cell lineage markers (lin–). While CFC assays and LTC-IC assays have proven useful in identifying the functional properties of flow cytometric sorted human cells, in vivo testing of selected cell populations in human subjects has not been possible for ethical reasons. Nonetheless, CD34+ selected hematopoietic cells have been proven to support patients long term after bone marrow transplantation (Verfaillie, 2002).
In an attempt to develop an in vivo system for detection of human hematopoietic stem cells, several groups have developed xenotransplantation models. Enriched populations of human hematopoietic stem cells from human bone marrow, mobilized peripheral blood, cord blood, or fetal liver have been observed to engraft in the preimmune sheep fetus (<60 days of gestation) with long-term evidence of multilineage peripheral human blood cell chimerism in the transplanted animals (Zanjani et al., 1992; Srour et al., 1993; Civin et al., 1996). Human cells present in the marrow of transplanted sheep fetuses can engraft in secondarily transplanted preimmune fetuses in vivo suggesting the presence of self-renewing stem cell populations.
Nonobese diabetic (NOD) mice bred with severe combined immunodeficient (SCID) mice result in NOD/SCID mice that accept human hematopoietic grafts (Schultz et al., 1995). This NOD/SCID model has permitted calculation of the frequency of human repopulating cells present in a donor sample (Dick et al., 1997). Similarly, SCID mice can be implanted with fragments of human fetal thymus and bone that will survive (Heike et al., 1995). Sublethal irradiation of these mice will impair hematopoiesis in the implanted human tissue and permit engraftment of intravenously administered human cells following transplantation (Namikawa et al., 1990). Other immunodeficient mice models that permit human hematopoietic cell engraftment have also been reported; however, it remains unclear whether any of these assays can accurately identify the same hematopoietic stem cells that repopulate the blood system of transplanted human patients (Dao et al., 1999).
STEM/PROGENITOR CELL DIFFERENTIATION: STOCHASTIC OR INDUCTIVE
The above methods have permitted subfraction of the hematopoietic system into a hierarchy of functionally and phenotypically distinct stem and progenitor cell compartments. However, little is known of the mechanisms that permit differentiation of the stem cells into the committed progenitor cell pools and further development into mature blood cells. Controversy remains whether stem/progenitor cell commitment and differentiation is the result of a stochastic (purely random process intrinsic to the cell) or an instructive (signals outside the cells dictate or determine cell fate) process (Enver et al., 1998).
Growth factors have been largely implicated as modulators of hematopoietic progenitor cell fate in support of the instructive model of lineage commitment. Experiments using hematopoietic cell lines or primary hematopoietic progenitor cells have provided results that the presence of stem cell factor (SCF) or granulocyte colony-stimulating factor (G-CSF) promotes neutrophil differentiation while M-SCF stimulates macrophage differentiation from GM-CFC in vitro (Metcalf, 1998). When individual GM-CFC are isolated as single cells in vitro, allowed to divide, and the paired daughter cells are then separated and exposed to single growth factors, GM-CSF has been demonstrated to promote neutrophil differentiation, but M-CSF promotes macrophage differentiation (Metcalf and Burgess, 1982). These results would support an instructive model of lineage commitment whereby the presence of a particular growth factor dictates the fate of progenitor cell differentiation. However, alternative interpretations of these data are possible.
Provision of bipotent or multipotent progenitors with a single growth factor in vitro and deriving a single lineage of differentiated cells could be interpreted (as above) that the growth factor dictated the fate of the progenitors. The same result could also be interpreted as the growth factor simply promoted the growth and survival of a subpopulation of the progenitors that had already committed to that particular lineage. Evidence that growth factors may play a limited instructive role for progenitors has been provided in analyzing several knockout mouse models. Mice that fail to express GM-CSF are viable but die as young adult mice due to pulmonary failure as a result of surfactant protein accumulation due to faulty macrophage handling of the protein. These mutant mice have apparently normal numbers of GM-CFC and neutrophils despite the absence of GM-CSF (Stanley et al., 1994). Similarly, G-CSF null mice display severely reduced numbers of circulating neutrophils, but GM-CFC are present at near normal frequencies and some neutrophils persist (Lieschke et al., 1994). These results suggest that the absence of GM-CSF and G-CSF stimulation of the progenitors has no apparent effect on GM-CFC commitment to mature cell lineage differentiation. One must be cautious, however, since compensatory pathways are often invoked when one molecular pathway is interrupted in knockout mice.
An alternative approach to addressing the role of growth factors in determining cell fate can be achieved by misexpressing a growth factor receptor for one lineage into another lineage using gene therapy. Transducing primary erythroid cells with a retrovirus encoding the M-CSFR has demonstrated that addition of M-CSF to the transduced cells promotes erythroid CFC formation but does not cause a shift in differentiation to macrophages or monocytes (McArthur et al., 1994). Similarly, providing multipotent progenitors with constitutively active Epo-R and M-CSFR does not result in a preferential production of erythroid or macrophage progeny, respectively, but does promote the proliferation of the progenitor cells in vitro (Pharr et al., 1994).
As noted above, transplantation of the CLP into lethally irradiated hosts demonstrates the restricted differentiation potential of these cells into T and B lymphocytes, natural killer cells, and antigen-presenting dendritic cells. However, introduction of the GM-CSFR into the CLP causes these cells to differentiate into granulocytes and macrophages with diminished lymphoid commitment (Kondo et al., 2000). Culturing CLP with IL-7 for several days in vitro causes these progenitors to become resistant to the signaling effects of an introduced GM-CSFR, and no differentiation into myeloid cells occurs. Most of the IL-7-stimulated CLP commit to B lymphoid development. Isolation of primary pro-B cells and introduction of the GM-CSFR reveal that these progenitors are resistant to the GM-CSFR transduced signals (Kondo et al., 2000). Early pro-T cells from the thymus can be reprogrammed for a myeloid outcome via introduction of the GM-CSFR and GM-CSF stimulation. These results suggest that during normal lymphocyte lineage specification, a phase of differentiation exists that only results in a lymphoid type outcome; however, the genome remains sufficiently plastic to permit rescue of another blood cell phenotype given certain conditions (Weissman et al., 2001).
Creation of transgenic animals carrying modified growth factor receptors has also been utilized to examine the role of growth factor signaling in lineage commitment. One example of such work was the creation of a transgenic mouse expressing a mutant growth factor receptor in which the cytoplasmic domain of the c-mpl receptor (receptor for thrombopoietin) was replaced with the cytoplasmic domain of the G-CSFR (Stoffel et al., 1999). The anticipated result was that transgenic mice homozygous for the mutant receptor would demonstrate deficiencies in platelet production similar to those of mice homozygous deficient for the c-mpl receptor. However, platelet production was relatively unaffected, suggesting that the mutant receptor was sufficient to mediate the intracellular signaling required for megakaryocyte proliferation and differentiation. These results support a stochastic lineage commitment process. The effects of a particular growth factor on the differentiation of a progenitor population depend on the relative expression of the relevant receptor for that growth factor on the progenitor cells. May and Enver (2001) have argued that the overall pattern of growth factor receptors expressed by a population of progenitor cells is likely a consequence rather than the determining cause of progenitor cell commitment. This conclusion begs the question of how the pattern of growth factor receptor expression is dictated in the various progenitor cell populations.
Transcription factors have been postulated to play a role in the process of cellular commitment to lineage differentiation. The SCL/tal-1 transcription factor is a basic helix-loop-helix protein that is expressed primarily in hematopoietic, endothelial, and central nervous cells (Shivdasanl et al., 1995; Robb and Begley, 1997). Knockout of this gene in mice results in embryonic lethality with a complete absence of hematopoietic cells (Shivdasanl et al., 1995; Robb et al., 1996). This deficiency most likely occurs at the level of mesoderm cell commitment to the hematopoietic lineage and is upstream of the hematopoietic stem cell. PU.1 is a protein expressed exclusively in hematopoietic cells (Fisher and Scott, 1998). Disruption of this molecule blocks both lymphoid and myeloid cell development (McKercher et al., 1996). Fetal liver hematopoietic stem cells isolated from homozygous null PU.1 mice failed to reconstitute lymphoid or myeloid lineages in lethally irradiated mice, although erythroid cells were derived from the mutant donor cells. Production of chimeric embryos via injection of blastocysts with PU.1 null embryonic stem (ES) cells resulted in viable mice that lacked evidence of PU.1 null ES cell contribution to B or T cells, monocytes, or neutrophils (Scott et al., 1997). GM-CSFR and G-CSFR are expressed normally in the absence of PU.1, but M-CSFR is not expressed. PU.1 null progenitor cells are not responsive to GM-CSF, G-CSF, or M-CSF. Transduction of PU.1 null progenitor cells with the M-CSFR overcomes the block in M-CSF-induced cell proliferation but does not restore full myeloid cell differentiation to these progenitor cells (Olson et al., 1995; DeKoter et al., 1998). Further studies concluded that PU.1 controls cellular differentiation by regulating distinct proliferation and differentiation pathways.
The role of transcription factors in controlling lineage specification can be addressed using other genetic approaches. Heyworth et al. (2002) recently transduced primary murine GM-CFC with a retroviral construct encoding a transcription factor (GATA-1) in an inducible form. They reported the progenitors differentiated into neutrophils and monocyte/macrophages in the presence of interleukin-3 (IL-3) or IL-3 plus Epo. In the presence of the retroviral vector alone or the inducible GATA-1 construct, transduced GM-CFC stimulated with the growth factors also differentiated into granulocytes only. However, induction of GATA-1 expression in the GM-CFC resulted in a dramatic appearance of erythroblasts, eosinophils, and basophils. This lineage switch was a high-frequency event occurring in >95% of individual clones of transduced GM-CFC. The induction of erythroid cells, eosinophils, and basophilic neutrophils occurred at the primary expense of monocyte differentiation in these experiments. These results suggest that a single transcription factor can reprogram the fate of committed GM-CFC.
How might the process of lineage specification be identified at the molecular level in multipotent cells? Recent evidence suggests that transcription of a number of lineage-affiliated effector genes, growth factor receptors, and transcription factors can be concomitantly identified in multipotent cells cultured in vitro (Hu et al., 1997). Similar evidence of simultaneous transcription of multiple lineage affector genes in single sorted stem and progenitor cells from mice and humans suggests that the low-level heterogenous expression of these molecules may provide potential starting points for lineage specification (Cheng et al., 1996; Delassus et al., 1999). For example, accessibility of gene transcripts for multiple lineage commitment pathways may lead to interactive or competing interactions leading to an upregulation (or downregulation of an alternative pathway) of appropriate molecules as a dominant commitment pathway emerges. In the future, gene expression profiling of stem and progenitor cells at different stages of normal or experimentally induced differentiation may permit elucidation of specific molecules associated with each state of commitment to lineage specification.
The primary sites of hematopoiesis change during mammalian development. Blood cells are first detectable in the mouse on embryonic day 7 (E7.0) and on the 18th day postconception in the human (Palis and Yoder, 2001). In both species, blood cells appear first in the yolk sac. The first murine hematopoietic cells are predominantly erythroid cells with few macrophages (Palis et al., 1999). Limited numbers of megakaryocytes are also produced within 12 hr of the first erythroid cells in the yolk sac. The first erythroid progenitors are called primitive erythroid progenitors (EryP) since these cells produce primitive erythroblasts. The primitive erythroblasts are large nucleated erythrocytes that contain fetal and adult hemoglobins. These cells are formed exclusively in the yolk sac from E7.0–E8.5 (Palis et al., 1999). The earliest macrophages are also unique in their cell surface protein expression and phagocytic function and are often referred to as primitive macrophages (Lichanska and Hume, 2000). Recently, a unique population of primitive megakaryocytes has also been identified. These cells demonstrate accelerated maturation of platelet formation compared to megakaryocytes produced later in development. In sum, the primitive erythroblasts, macrophages, and megakaryocytes comprise the primitive hematopoietic phase of blood cell development.
The first definitive hematopoietic progenitor cells are also produced in the murine yolk sac on E8.25 (Palis et al., 2001). Of interest, these definitive progenitor cells do not normally differentiate into mature blood cells in the yolk sac, but rather enter the bloodstream at the time the liver is beginning to accumulate hematopoietic progenitors. Thus, yolk sac-derived definitive progenitor cells likely seed the liver where they subsequently mature into circulating granulocytes and red blood cells.
Are the primitive and definitive hematopoietic progenitors derived from a common stem cell? The most evidence of such a relationship has been determined from experiments using murine ES cell differentiation in vitro (Kennedy et al., 1997). ES cells expressing flk-1 can give rise to primitive and definitive hematopoietic progenitor cells. This question has not been answered in murine embryos, but the controversy has been reviewed extensively (Cumano and Godin, 2001; Yoder and Palis, 2001). Current evidence suggests that primitive hematopoiesis may be derived via mechanisms distinct from definitive hematopoiesis.
Available data suggest that stem cells derived from the aorta-gonad-mesonephros (AGM) region may be the first to seed the fetal liver (de Bruijn et al., 2002; North et al., 2002). Recent studies suggest that wherever the stem cells originate from, these cells are expanded locally in the AGM and within the yolk sac circulation concomitant with proliferation of the stem cells in the early fetal liver to account for the dramatic increase in stem cell numbers during the first 48 hr of liver hematopoiesis (Kumaravelu et al., 2002). The liver continues as the primary hematopoietic site until late in gestation when stem cells reenter the circulation and seed the marrow and spleen.
The spleen serves as an active hematopoietic site throughout the life of the mouse but is principally a lymphoid organ in humans. While stem cells can be identified in the adult murine liver, recent data suggest that a certain fraction of marrow-derived stem cells are constantly circulating throughout the body and thus may contaminate all perfused organs (Wright et al., 2001). In general, the murine liver loses the ability to support hematopoiesis within a few weeks of postnatal life but can regain such function in the face of extreme hematopoietic stress.
STEM CELL PLASTICITY
Stem cells residing in adult tissues have typically been thought to be committed to the production of progeny specific for that particular tissue. Thus, hematopoietic stem cells give rise to blood cells, not liver cells, neurons, or muscle cells. Or do they? Numerous publications over the past 3–4 years have provided some evidence for more plasticity of stem cell commitment and differentiation than previously considered possible. Bone marrow cells, including isolated populations of hematopoietic stem cells, have been reported to differentiate into muscle, neurons, liver, vascular, lung, intestine, and kidney tissue (Gussoni et al., 1999; Jackson et al., 1999; Brazelton et al., 2000; Lagasse et al., 2000; Krause et al., 2001; Grompe, 2002). These results have created significant controversy, particularly with regard to the criteria used to demonstrate that a specific stem cell directly differentiated into an unexpected lineage. Recent publications point out that some of the results of stem cell plasticity studies may be explained due to cell fusion events or occur at such low frequency that the progeny of the stem cells would not contribute to tissue function (Iscove, 2001; Weissman et al., 2001; Wagers et al., 2002; Ying et al., 2002).
Several investigators have suggested some basic principles that should be addressed in future stem cell studies (Anderson et al., 2001; Weissman et al., 2001; Graf, 2002; Lemischka, 2002). First, the stem cell population should be localized to a specific cell phenotype that can be isolated and therefore comparable from one study to the next. The contribution of the stem cell to a particular organ should be clonal. That is, one should prove that the differentiated tissues derived from a donor source arose from single stem cells. The tissue derived from the stem cell should be a significant population of cells that demonstrate functions consistent with the organ of residence. Following these simple principles should clarify the question of whether an adult tissue stem cell can be stimulated to commit to differentiation to a lineage of cells of a different tissue upon transplantation.