Based on historical radiation experiments in rodents, the hematopoietic stem cell was defined by its biological properties and later by the expression of certain surface antigens (e.g., CD34), as well as the absence of lineage-specific markers (e.g., DR). Quite recently it was shown that hematopoietic reconstitution can also be achieved by CD34− stem cells, which can be isolated from the bone marrow, peripheral blood and cord blood cells. CD34− stem cells are considered to be predominately part of the quiescent stem cell pool of hematopoietic and mesenchymal stem cells. Due to novel techniques, CD34− stem cells can be expanded on the level of a true stem cell but also directed towards their differentiation into specified tissues or organ systems. This requires the establishment of primary fibroblast-like CD34− stem cells in vitro and their possible reversible and transient immortalization with optimized vector systems.
First in 1949, Jacobsen et al.  and in 1951 Lorenz et al.  showed the possibility of injecting bone marrow intravenously to rescue lethally irradiated animals. In 1961, Till and McCulloch showed that spleen-derived hematopoietic stem cells (HSC) were also able to reconstitute an irradiation-induced hematopoietic failure in mice . They found further that each spleen colony was derived from a single clonogenic precursor which gave rise to cells of all lineages. Therefore, at that time, HSC were defined as: A) having the capacity for radioprotection; B) being able to develop cells of all hematopoietic lineages, and C) having the capacity for self-renewal.
Soon it became clear that HSC were only a fraction of hematopoietic cells which could be harvested from the spleen or bone marrow [4, 5]. Therefore, hematopoietic cells were subfractionated based on size, density, and the expression of certain cell surface markers [6-10]. An HSC was first defined in the mouse by the expression of the stem cell antigen (SCA-1+), the low expression of the Thy-1 marker (Thy-1low), and the complete absence of any hematopoietic lineage markers (lin–) [8, 11, 12]. These cells represent approximately 0.05% of the marrow cells (1 in 2,000 marrow cells) . This definition was later transferred to the human system, where the HSC is currently defined and used as CD34+DR–lin–. But it is still controversial whether the earliest identifiable HSC is DR– or DR+ . Huang and Terstappen reported that a CD34+DR– pluripotent stem cell could give rise to CD34+DR+lin– HSC and a mixed population containing CD34+DR+lin– HSC and DR– stromal elements . Nevertheless, the expression of DR on an HSC seems to be one of the first steps of hematopoietic differentiation [16, 17]. But a DR+ HSC is still able to develop cells of all lineages and is therefore still considered to be a true stem cell . This was confirmed by Srour et al., who reported that CD34+DR+ marrow cells give rise to more differentiated precursors, while high proliferative potential colony-forming cells (HPP-CFC) are contained predominately among CD34+DR+ progenitors ; the HPP-CFC could be further enriched by selecting for c-kit+ cells which express the receptor for stem cell factor (SCF) . One reason why even early hematopoietic progenitors express major histocompatibility complex class II molecules might be a mechanism for the induction of self-tolerance during ontogeny . Although it was shown that hematopoietic progenitors are of clonal origin, it still remains controversial whether there is one absolute stem cell which undergoes self-renewal with every cycle (“spermatocyte model”) or the number of HSC is limited and maturation to hematopoietic precursor cells occurs consecutively (“oocyte model”).
The definition of the HSC was extended to in vivo models, where different marrow cells were transplanted into lethally irradiated mice to reconstitute the myeloablated bone marrow. A marrow transplantation from the initially transplanted mice serving as donors had to reconstitute a second generation of mice. The cells responsible for such a long-term marrow reconstitution were termed severe combined immunodeficiency “(SCID) repopulating cells” (SRC), and were also defined in vitro as “long-term culture-initiating cells” [22, 23].
The stage of differentiation and maturation of an HSC and the hematopoietic progenitors also defines their responsiveness to growth factors. This is important for the in vivo and ex vivo expansion of HSC, which will be discussed later. The differentiation of a pluripotent stem cell to a lineage-committed or even mature blood cell underlies the response to a variety of growth factors [24, 25], of which SCF (c-kit ligand) has an exceptional role [26, 27]. SCF is produced mainly by marrow stromal cells [28, 29] and HSC express the SCF-receptor [30, 31], which is c-kit, a tyrosine kinase [32, 33]. SCF itself has an important role in the survival of HSC with regard to self-renewal and as a comitogen in the movement of HSC out of the HSC pool into the progeny pool [34, 35]. HSC usually do not respond to any given cytokine singly (even SCF alone only seems to provide long-term survival of nondividing HSC) , but the addition of other growth factors in a double, triple or multiple combination directs the destiny of now committed progenitors . The application of SCF in combination with interleukin 1 (IL-1), IL-3 or IL-6 causes a significant fraction of HSC to proliferate along the myeloid or erythroid lineages; the addition of G-CSF, GM-CSF or macrophage colony-stimulating factor (M-CSF) directs the HSC mainly to differentiate into myeloid precursors . One of the most effective combinations in vitro was the treatment of CD34+ cells with six growth factors (IL-1, IL-3, IL-6, G-CSF, GM-CSF and SCF) .
SOURCES OF HSC
Traditionally, stem cell transplantations were performed by the infusion of an unmanipulated, complete mixture of marrow cells [40-42]. The marrow suspension consists of a complex heterogeneity of cells, including the essential population of HSC. With increasing disparity between donor and recipient, in the case of an allogeneic transplant, severe side effects such as graft-versus-host-disease (GVHD) emerged. The attempt to deplete marrow grafts of T cells substantially reduced the frequency and severity of GVHD, but also occasionally resulted in poor engraftment [43, 44]. It became obvious that a stem cell transplant with complete marrow yielded a lower rate of graft failure and a significantly lower relapse rate. Other purging methods include the use of monoclonal antibodies directed against various B-cell epitopes, as used in autologous transplants for patients with non-Hodgkin's lymphoma (NHL) or multiple myeloma [45, 46]. A stem cell transplant with a complete marrow cell suspension requires the infusion of a fairly large amount of cells, which should be in the range of 0.4-1.2 × 108 mononuclear cells (MNC)/kg body weight . Nevertheless, alternative approaches became desirable. They are based on the positive selection of cells with long-term repopulating capability . In humans, these cells are still defined as CD34+lin– and have a high probability of being free of contaminating malignant cells in autografts.
Peripheral Blood Stem Cells (PBSC)
Although bone marrow has the highest percentage of CD34+ MNC per ml, an even higher absolute number of CD34+ cells can be collected from peripheral blood after mobilization with chemotherapy and growth factor. The large amount of peripheral blood which can be processed during apheresis yields this good result . That peripheral blood could be used to reconstitute lethally irradiated animals was originally demonstrated in the rat  and later in the dog model . Peripheral blood which is not mobilized contains a very low percentage of CD34+ MNC (approximately 0.15%), but the mobilization with a short course of chemotherapy in combination with growth factor treatment increases the percentage of CD34+ cells [49, 52-55]. Another advantage of using peripheral blood is the fact that it is less likely to contain malignant cells compared with marrow [56, 57].
The isolation of PBSC is based on the definition of the cell which most likely has long-term repopulating capability. Due to the lack of in vitro systems which prove the existence of such a population before it is transplanted into the recipient, every approach to date is still a compromise. Peripheral blood MNC (PBMNC) are considered to contain HSC. MNC collected during leukapheresis have been shown to contain colony-forming units granulocyte macrophage (CFU-GM) and CD34+ cells [58, 59].
CD34 is a transmembrane cell surface sialomucin which is expressed on hematopoietic progenitor cells [60-62] and on vascular endothelial cells [63, 64]. The CD34 antigen is widely expressed on the pluripotent hematopoietic progenitors, but not on the neoplastic cells of patients with NHL, myeloma and most solid tumors. Mobilized PBMNC can be harvested and incubated with a biotinylated anti-CD34 antibody. The coated cells can be applied to a column of avidin-coated beads. The CD34+ cells can then be isolated and cryopreserved.
During fetal development the liver is physiologically a part of the hematopoietic tissues. From the second to the seventh months of pregnancy and ideally before the onset of lymphopoiesis, the fetal liver can be used for transplantation. It has been shown that fetal liver cells can reconstitute successfully both hematopoietic and lymphopoietic systems as shown clinically in children with congenital immunodeficiencies [65, 66].
Umbilical Cord Blood
During hematopoietic ontogeny and the perinatal shift of the sites for hematopoiesis, HSC reach the blood and circulate. The advantage of cord blood is that it can be assessed at the time of birth without affecting the fetus or the mother. It has been shown that with refined techniques, up to 200 ml of cord blood can be obtained, containing as many as 4 × 106 myeloid progenitors . These cells may be used, even after cryopreservation for successful transplants in even older siblings or as target cells for a gene transfer [68-70].
IN VITRO EXPANSION OF STEM CELLS
The expansion of hematopoietic progenitors is used to supplement a standard autograft. This technique will provide more mature and functional neutrophils, which should decrease the period of neutropenia even further than the in vivo application of growth factors . Certain combinations of growth factors can increase the CFU-GM production in vitro by 20-200-fold on day 7 . The induction of differentiation is also of interest in conjunction with gene transfer experiments. This might not only increase the transduction efficiency, but also provide sustained serum levels of genetically transferred constructs. On the other hand, differentiation can also be a problem, since true stem cells have to be maintained in vitro. This might be solved by the use of transiently immortalized CD34– HSC (see below).
GENE TRANSFER INTO HSC
HSC are attractive targets for gene transfer because of their ability to provide long-term hematopoiesis. Furthermore, gene transfer into HSC would lead to a continuous supply of genetically modified hematopoietic cells for the lifetime of the recipient. The advantage of retroviral vectors is the ability to stably integrate genes efficiently into the genome of target cells. However, efficient and safe retrovirus-mediated gene transfer was not possible until the development of packaging cells which allow the production of replication-defective recombinant virus, free of wild-type virus . Major disadvantages of moloney murine leukemia virus-derived retroviral vectors include the inability to infect nondividing cells and the requirement for the appropriate amphotropic receptor on the target cell membrane . Another tool for the transfer of genes into target cells are the adeno-associated viruses, which have been shown to allow the production of high virion titers, stable recombinant virions without helper contamination and with potentially better safety parameters [75-77].
HSC have therefore been difficult to transduce. Recently, improved marrow culture conditions and the identification of growth factors that promote the replication of HSC have increased the gene transfer efficiency into these cells. But instead, CD34– stem cells present a fibroblast-like morphology. Being capable of transiently immortalizing these predominately quiescent cell types, HSC will generate highly efficient targets for gene therapy.
CD34– STEM CELLS AND THEIR BIOLOGY
During ontogeny, hematopoiesis occurs in the yolk sac and then shifts to spleen and liver and eventually to the marrow cavity. In the bone marrow, differentiation and proliferation of HSC generally depend upon an intact “microenvironment.” The direction of differentiation is determined by different parameters, such as concentration and composition of interacting growth factors [78, 79]. Stromal cells are thought to be the major source of those growth factors and also interact with HSC on an intercellular level . Both HSC and stromal cells are mesoderm-derived and were considered to be two different entities. Nevertheless, Singer et al. described adherent common precursors for stromal and hematopoietic cells [81, 82]. Huang and Terstappen suggested that a single fetal CD34+ DR– CD38– stem cell differentiates into stromal elements and cells with hematopoietic characteristics , although they had to correct their conclusions . The “stromal cell” versus “stem cell” dichotomy raised for the first time the discussion on the phenotype of the earliest HSC population. Over many years, it was thought that stromal cells are CD34–, while even quiescent stem cells express at least low levels of the CD34 antigen. But recently, different groups demonstrated independently that hematopoietic reconstitution is possible with CD34– cells in various animal models, using bone marrow-derived HSC [83-87]. Nevertheless, this subject still remains controversial. Other researches find that hematopoietic reconstitution still requires CD34+ cells, while CD34– cells facilitate repopulation in the nonobese diabetic (NOD)/SCID environment .
Nevertheless, marrow stroma-derived fibroblast-like CD34– cells can give rise to CD34+ cells with hematopoietic characteristics with regard to colony formation and long-term-culture initiation . Those HSC can be shifted either towards differentiation or proliferation by different growth factors or culture conditions : while the ligand for the tyrosine-kinase receptor c-kit SCF induces differentiation towards committed hematopoietic progeny , IL-6 rather promotes proliferation of CD34–, adherent growing HSC. IL-6 is also capable of reversing the transition from fibroblast-like adherent growing CD34– HSC to more committed CD34+ HSC, an ability we took advantage of to generate CD34–, adherent growing HSC from human PBMNC . These fibroblast-like cells can again spontaneously differentiate into CD34+ nonadherent cells, which, though, is dependent upon the autocrine and paracrine production of SCF .
CD34– HSC were also isolated from small amounts of murine peripheral blood cells and cloned after immortalization, using an SV-40 large T-antigen containing retroviral vector. Recipient mice were sucessfully transplanted with a single stem cell clone after lethal irradiation . The morphology of this CD34– stem cell clone showed a fibroblast-like morphology and the phenotype revealed a coexpression of Sca-1 and c-kit (CD117). Thy-1 was also expressed on a low level. This set of experiments showed that CD34– HSC present the morphology of stromal-like cells but this phenotype is such a true HSC. Interestingly, using CD34– HSC clones for hematopoietic reconstitution, additional treatment with growth factors did not accelerate hematopoietic recovery, suggesting the necessary factors are produced by this kind of cells in an autocrine or paracrine fashion . The analysis of bone marrow biopsies from recipient mice showed that transplanted cells with the known phenotype and fibroblast-like morphology were located as bone-lining cells along the bone spicules in the marrow. This is usually considered the niche for osteoblastic cells (see below).
Also human bone marrow stroma-derived CD34– fibroblast-like cells were analyzed for their potential to differentiate into hematopoietic progeny. These cells produced CFU and engrafted in NOD/SCID mice. Cells also expressed c-kit (CD117) and mesenchymal markers such as osteocalcin.
Figure 1 shows the possible horizontal shift between a commonly known “stromal cell,” a quiescent stem cell and an active stem cell. Stroma-derived growth factors affect the active stem cell as well as the more committed HSC and the mesenchymal stem cell.
The “Stem Cell Cycle”
It was demonstrated that a monoclonal CD34– stem cell line is able to reconstitute all hematopoietic lineages . Using CD34– fibroblast-like stem cells for intravenous infusion, hematopoietic recovery of the recipient mice was not delayed in any part after myeloablative, total body irradiation. This suggests that circulating stem cells also might even initially return to their “place” within the marrow microenvironment before differentiating into committed hematopoietic progenitors. We could demonstrate that CD34– fibroblast-like HSC contain hematopoietic pluripotency and a certain number of those cells circulate in the peripheral blood. They can still return to their settling environment within the marrow stroma. We call this the “stem cell cycle” . The “stem cell cycle” defines the circulation from CD34– stem cells resting within the marrow microenvironment until they eventually differentiate and circulate in the peripheral blood. Circulating CD34+ cells, though, can return to the bone marrow and downregulate CD34– expression, while growing as fibroblast-like cells. Using immortalized CD34– stem cells, clones can be used to participate in the “stem cell cycle” and achieve complete hematopoietic reconstitution with genetically modified stem cells (Fig. 2).
Hematopoietic reconstitution, though, does not occur in terms of a linear dimension. As shown in animal reconstitution experiments, transplanted cells show their offspring population in the periphery in a timely manner. For some time, the transplanted cell type is present in the periphery and vanishes again, before it returns again some time later. Morley et al. called this the “oscillatory nature of hemopoiesis”  and postulate a “bone marrow transit time” for stem cells within the “stem cell cycle.” The “marrow transition time” is dependent upon a phase of quiescence among the CD34– stem cell population. The vast majority of CD34– stem cells are quiescent while they rest within the marrow microenvironment. This quiescence is mediated by cell cycle-associated regulatory mechanisms and the lack of the activation of signal transduction pathways . Other possibilities to maintain a state of quiescence include the lack of expressing growth factor receptors [unpublished]. The mechanism of quiescence of the majority of HSC in the marrow pool is an efficient mechanism to provide a life-long pool of hematopoietic progeny. Only a limited number of activated HSC can respond to growth factor-mediated signals and provide a sufficient number of cells to fight infections or cancer.
ISOLATION OF CD34– STEM CELLS
Taking advantage of the postulated “stem cell cycle” and the homing capabilities of circulating stem cells, CD34– fibroblast-like progenitor cells can be isolated from peripheral blood by cytokine-mediated adherence to the plastic surface of tissue culture flasks. One growth factor that can facilitate the adherent growth and maintains the fibroblast-like morphology is IL-6. This factor is also a strong inducer of proliferation of CD34– stem cells and inhibits differentiation into more committed hematopoietic progenitor cells [89, 90]. Although it is possible to establish fibroblast-like cells from small amounts of PBMNC in vitro , there is a crucial cell density per well or flask to provide necessary cell-cell interaction and sufficient concentrations of paracrine growth factors. The fibroblast-like cells also produce a wide variety of growth factors, as demonstrated for bone marrow stromal cells . Nevertheless, the life span of CD34– fibroblast-like cells is limited in vitro. But those cells can be immortalized by various constructs  and cell clones can be established . These clones can be analyzed with regard to functional properties and their ability to achieve long-term hematopoietic reconstitution. Using single clones for engraftment obviously does not hamper complete hematopoietic reconstitution in animals , but instead fibroblast-like CD34– stem cells can be expanded almost indefinitely, maintaining a foremost immature level of differentiation. Currently, vector systems for transient and reversible immortalization of CD34– stem cells are being optimized (unpublished).
CD34– fibroblast-like progenitors could be isolated already from the bone marrow, as well as from peripheral blood [92, 94]. But preliminary data show that this cell population can also be isolated very efficiently from cord blood cells (unpublished). Although MNC are a potential source of CD34– fibroblast-like stem cells, it may be of advantageous to use CD34+ precursor cells after stem cell mobilization to start with an enriched population generating large numbers of adherent cells in vitro.
MESENCHYMAL STEM CELLS
There is still an ongoing discussion whether there is a common precursor cell of the marrow microenvironment and hematopoiesis [99, 100]. There is also more and more evidence that there are CD34– resting or quiescent HSC along with other fibroblast-like cells which serve as bystander cells and provide necessary growth factors and even cell-cell contact to equilibrate the sensitive balance of differentiation and proliferation of hematopoietic progeny . There is also growing evidence that so-called “mesenchymal stem cells” actually reconstitute the marrow stroma and release committed progenitor cells into the circulation , basically participating in the “stem cell cycle.” Mesenchymal stem cells, though, are precursor cells of other mesenchymal organ systems, such as chondrocytes, osteoblasts, and myoblasts . First clinical trials were already performed using bone marrow-derived mesenchymal cells to treat children with osteogenesis imperfecta . But CD34– mesenchymal stem cells are also apparently able to generate more specified tissue, such as endothelial cells and even cardiomyocytes . Since the CD34 antigen is lacking early hematopoietic and mesenchymal stem cells, major efforts were made to identify novel markers for this type of cell. Ziegler et al. found the KDR receptor on CD34– HSC , suggesting also the potential for neoangiogenesis.
There is obviously a common progenitor cell of CD34– hematopoietic and mesenchymal stem cells. This type of cell can give rise to various specified tissues (Fig. 3), depending on growth factor-mediated signals and an internal signal control. Also using novel vector systems for quiescent cells , CD34– hematopoietic and mesenchymal stem cells can be used very efficiently for cell and gene therapy with a wide spectrum of applications.
Entering a new millemium, stem cell transplantation has become an increasingly important clinical modality for the treatment of a wide variety of onco-hematological and metabolic disorders. With the improvement of therapeutic supplementations, the incidence of severe and fatal side effects declined, and even unrelated and mismatched transplantations are performed on a routine basis. Nevertheless, it is still desirable to minimize any kind of complications. The characterization and the handling of CD34– stem cells will contribute to a new development that will broaden the indications for cellular and gene therapy. With the extended knowledge on the biology of hematopoietic and mesenchymal stem cells, it will be possible to develop new treatment strategies and we will be getting closer to generating whole organ systems in vitro, derived from single autologous blood cells.
Nevertheless, there are concerns by hematologists who had taken CD34+ HSC from the bench to the bedside, applying them very successfully in clinical practice. But as pointed out by Margaret Goodell , CD34– stem cells will probably not entirely replace the stem cell transplantation with CD34+ hematopoietic progenitors, but will provide a new tool in the treatment of diseases which affect the hematopoietic and mesenchymal organ system. CD34– stem cells can be isolated from their major source, the bone marrow, but due to the circulation of stem cells, CD34– HSC can also be generated from even small amounts of peripheral blood and cord blood cells.
This work was supported by the Deutsche Forschungsgemeinschaft, Bonn-Bad Godesberg, Germany.