|Home | About | Journals | Submit | Contact Us | Français|
Hematopoietic stem cells (HSC) represent the prototype stem cell within the body. Since their discovery, HSC have been the focus of intensive research, and have proven invaluable clinically to restore hematopoiesis following inadvertent radiation exposure and following radio/chemotherapy to eliminate hematologic tumors. While they were originally discovered in the bone marrow, HSC can also be isolated from umbilical cord blood and can be “mobilized” peripheral blood, making them readily available in relatively large quantities. While their ability to repopulate the entire hematopoietic system would already guarantee HSC a valuable place in regenerative medicine, the finding that hematopoietic chimerism can induce immunological tolerance to solid organs and correct autoimmune diseases has dramatically broadened their clinical utility. The demonstration that these cells, through a variety of mechanisms, can also promote repair/regeneration of non-hematopoietic tissues as diverse as liver, heart, and brain has further increased their clinical value. The goal of this review is to provide the reader with a brief glimpse into the remarkable potential HSC possess, and to highlight their tremendous value as therapeutics in regenerative medicine.
When scientists discovered that a rare population of stem cells present within the bone marrow maintains the entire adult hematopoietic system, the door was opened to the possibility of applying regenerative medicine to the field of hematology.
The diverse array of hematopoietic cells circulating within the peripheral blood at any given point is produced within the bone marrow through the exquisitely coordinated process of hematopoiesis. Remarkably, under steady-state conditions, this process generates roughly ten billion red blood cells and one billion white blood cells every hour of every day during the lifetime of a human, while simultaneously ensuring that the levels of each cell lineage within the circulation are maintained within precisely regulated limits. In addition to supplying all these cells during steady state, the bone marrow rapidly responds to demands, increasing the production of specific blood cells to counter a variety of insults and injuries, thus ensuring homeostasis is quickly restored.
Adult hematopoiesis is supported by rare hematopoietic stem cells (HSC) that account for only about 0.01% of the total cells in the bone marrow . The complex hematopoietic process is orchestrated by the bone marrow microenvironment, which provides a network of extracellular matrix, vasculature, and stromal cells that supply the support/factors necessary for HSC survival, division, and differentiation. There are two distinct HSC niches within the marrow, the osteoblastic and the perivascular. A good deal of controversy exists as to whether HSC selectively reside in one or the other of these niches, and a great deal of interest is currently focused on interrogating the perivascular niche, which is frequently located close to trabecular bone, and is comprised of mesenchymal stromal cells (MSC) and endothelial cells [2–10].
Although many reviews of the HSC field do not make mention of it, it is important to note that some of the first compelling evidence that stem/progenitor are present in the bone marrow, and that damage to the chromatin of these primitive cells could explain the pronounced effects radiation exerted upon hematopoiesis, came from studies performed as early as the 1930s by Florence Sabin [11, 12]. Later, in the 1950s, Nobel laureate E. Donnall Thomas conducted pioneering work on identical twins that clearly demonstrated that primitive stem-like hematopoietic cells were present within the bone marrow, and that these cells had the ability to restore hematopoiesis after lethal irradiation . However, it was the pivotal series of studies conducted by Till and McCulloch in the 1960s that provided the definitive proof for the existence of the multipotent, self-renewing, serially transplantable HSC within normal mouse hematopoietic tissue [14, 15]. Collectively, these studies established the clonal nature of hematopoiesis, and introduced the concept that transplanting a single multipotent HSC can regenerate the entire hematopoietic system within a suitable recipient.
The fact that the entire hematopoietic system can be repopulated following infusion of limited numbers of HSC makes it possible for HSC transplantation to treat/cure a range of hematologic diseases, including malignancies.
Following detonation of the first atomic bombs, physicians and scientists eagerly sought ways to protect the hematopoietic system from the life-threatening effects of ionizing radiation, and to develop treatments to restore this critical system following exposure. Jacobson and colleagues were the first to report that shielding the spleen or femur of mice with lead protected the mice from the lethal effects of ionizing radiation on blood production. Interestingly, at the time, the observed protection was interpreted to be due to the production of some unknown factor within the spleen and marrow that stimulated recovery/regeneration of the hematopoietic system . However, a series of studies performed by Lorenz et al., Barnes and Loutit, and Main and Prehn [17–23] rapidly disproved this mistaken notion by showing that radio-protection could also be conferred by intravenous infusions of bone marrow, and by providing definitive evidence that the observed radio-protection was, in fact, due to HSC that were present within the infused marrow. These groundbreaking observations began the era of HSC transplantation, bringing the hope/dream that this new treatment could ultimately cure patients with life-threatening hematological malignancies by allowing physicians to escalate radiation doses to levels that would completely ablate the marrow, and thus ensure killing of all of the cancer cells, and to then rescue the patient by infusing bone marrow cells from a healthy donor to restore normal blood production.
A number of successes were quickly achieved in mouse models of bone marrow transplantation (BMT), which created a great deal of optimism in the field, and quickly led to premature attempts to translate this promising new therapy into the clinic. Thomas and colleagues at the Columbia University-affiliated Mary Imogene Bassett Hospital (and later at the Fred Hutchinson Cancer Research Center (FHCRC)) spearheaded many of these early efforts. They began their human HSC transplant program in the 1950s, and quickly proved that fairly large volumes of bone marrow could safely be infused into the circulation of human patients, and subsequently produce hematopoietic engraftment/repopulation, although this engraftment was transient in nature . Within only a few years, this same group reported on the treatment of two acute lymphoblastic leukemia (ALL) patients, using high-dose total body irradiation (TBI) to kill the leukemic cells, followed by an infusion of bone marrow from twin (syngeneic) siblings to rescue the hematopoietic system of these patients . Concurrently, the renowned French oncologist, George Mathé, performed the first European allogeneic BMTs, in the hopes of saving the lives of five Yugoslavian workers who had accidentally been exposed to high-dose radiation as a result of a Criticality accident at the Vinča Nuclear Institute. This report is particularly interesting, because four of the five recipients survived, despite exhibiting only partial and transient chimerism. In retrospect, it appears their survival was most likely due to limited numbers of surviving autologous HSC reconstituting hematopoiesis while under the support/protection of the infused allogeneic marrow cells [25–27]. In another contemporaneous study, McGovern et al.  successfully reconstituted a terminal ALL patient with autologous bone marrow cells that had been harvested during a previous remission and cryopreserved prior to infusion.
Collectively, these studies provided convincing proof that infusing bone marrow cells could protect against irradiation-induced bone marrow damage/destruction and mediate rapid clinical and hematological recovery. However, these studies also revealed that even supralethal doses of irradiation were not sufficient to eliminate hematologic malignancies, and results of most clinical BMT attempts in patients with leukemia and lymphomas were largely disappointing [25, 29]. These discouraging outcomes likely stem from the fact that all of the preclinical BMT studies were performed in inbred mice, which are genetically identical, and therefore require no immunological matching between donor and recipient . In contrast, with the exception of the study Thomas performed in twins, all other clinical BMTs had used cells isolated from allogeneic donors, which unknowingly introduced two major immunological challenges that do not exist in inbred mice: 1) the recipient’s immune system rejecting the grafted cells, which leads to graft failure; and 2) graft-versus-host disease (GVHD), in which host-reactive donor T lymphocytes attack the recipient’s skin, liver, and gastrointestinal tract, causing damage that, if left unchecked, can be fatal [31, 32]. These initial clinical failures led many investigators to conclude that the immunological barriers posed by transplanting bone marrow cells from one individual to another would prove insurmountable, and that BMT would not likely ever be used in clinical medicine to treat hematological malignancies .
Thomas’ group, however, was not dissuaded by these clinical setbacks, and they began work in earnest to learn how to overcome the host-recipient immune barrier. Their Nobel prize-winning studies in the dog model led to the development of new methods of bone marrow cell isolation and host immunosuppression, and novel agents to combat graft rejection. Perhaps most importantly, this group also developed a canine histocompatibility typing system, which enabled them to show that, to reduce the risks of both graft rejection and of GVHD, it was essential to have histocompatibility matching between donor and recipient [33–66]. These studies in the canine model led to a marked improvement in the understanding of the human major histocompatibility complex (HLA), making matching possible in human patients . They also revealed, however, that GVHD can still pose a serious problem for a well-matched donor and recipient, unless specific immunosuppressive agents to block T lymphocyte proliferation are administered [29, 39, 44, 57–62, 67–70].
Once it was recognized that GVHD was caused by T lymphocytes present within the HSC graft, multiple groups tested T lymphocyte depletion or immunoselection of HSC, and showed that both methods of graft manipulation markedly reduced the risk of GVHD following allogeneic HSC transplantation [71, 72]. Equally as exciting, investigators came to the realization that, by performing immunoselection to enrich for HSC, it was possible to use autologous HSC for hematological malignancies, because this immunoselection could result in a 30- to 50-fold reduction in tumor burden within the graft [73–76]. Unfortunately, however, it was soon discovered that these benefits came at a very high price, since grafts that had been T-depleted or immunoselected often engrafted less effectively/were rejected and, in the case of the immunoselected grafts, the time to reconstitution was significantly prolonged, placing patients at risk of significant hematologic morbidities and opportunistic infection [77–81]. Luckily, during this same time period, advances in infectious disease therapeutics provided new means of protecting the recipient during this prolonged period of vulnerability [82–84].
Somewhat unexpectedly, more detailed analyses of emerging preclinical and clinical data revealed that allogeneic HSC transplantation could also produce a second potent therapeutic effect — the so-called graft-versus-tumor (GVT) effect — in which donor lymphocytes present in the graft recognize and kill the host’s tumor cells. Barnes’ group showed that leukemic cells were eradicated when mice received allogeneic, but not syngeneic, BMT, demonstrating this important GVT effect for the first time [85, 86]. These investigators then had the insight to propose that cells within the graft could mount an immune response against the host leukemia, and might thus play a critical part in killing cancer cells. In 1965, Mathé achieved the first survival of an allogeneic bone marrow graft in a leukemia patient, and showed that the graft produced an anti-leukemic effect he termed “adoptive immunotherapy” [23, 25].
Disappointingly, as more transplant data became available, physicians recognized that patients who developed GVHD were much less likely to suffer cancer relapse [87–89], and it was soon revealed that GVHD and GVT were intimately linked. As a result, while T lymphocyte depletion and immunoselection of HSC could dramatically reduce the incidence of GVHD, these manipulations would inexorably remove the ability of the infused marrow cells to mediate the critically important GVT effect, profoundly increasing the patient’s risk of relapse [77, 80, 81, 90–95]. Similarly, Kolb and colleagues (in 1990) became the first to show that it was possible to successfully induce remission in 3 patients with relapsed chronic myelogenous leukemia (CML) following an allogeneic transplant, by infusing donor lymphocytes (DLI) . However, just as occurred transplants in which the initial graft contained high levels of T cells, two of these three patients developed GVHD following DLI.
Not surprisingly, the conventional HSC transplantation procedure using high-intensity radiation/chemotherapy to ablate the patient’s endogenous marrow, poses serious risks and side effects, restricting the therapy to relatively young patients who are in good medical condition. As such, this potentially life-saving therapy cannot be used to treat some of the most common hematological disorders, for which the median age at diagnosis ranges from 65–70 years . Fortunately, the discovery of the GVT effect made possible the development of the newer “reduced-intensity transplant” protocols, which do not destroy host HSC [97–103]. In contrast to a traditional allogeneic HSC transplant, this ingenious approach relies on the GVT effects of the graft, rather than the myeloablative effects of irradiation/chemotherapy, to kill cancer cells. Patients receive non-myeloablative irradiation, immunosuppressive drugs, and HSC transplantation, producing mixed hematopoietic chimerism. If the patient does not experience GVHD, he/she can then be infused with additional donor lymphocytes to promote conversion to completely donor-derived hematopoiesis, with the goal of inducing a robust GVT effect and complete eradication of the malignancy. The GVT effects are slow, with remission in some patients taking many months to achieve, but have produced very promising clinical results thus far. In addition, the lack of regimen-related toxicity now allows HSC transplantation to be performed in elderly/very sick patients who cannot be subjected to high levels of irradiation, and were thus ineligible for a transplant. Importantly, studies have now provided evidence that the GVT effect also makes it possible to combat metastatic tumors arising in other non-hematopoietic organs. Currently, researchers are focusing on better understanding the precise T lymphocyte populations responsible for GVHD and GVT, in the hopes of finding ways to induce a robust GVT effect without inducing GVHD.
Although HSC were first identified in and isolated from bone marrow, alternate sources of HSC such as mobilized peripheral blood and umbilical cord blood have many characteristics that make them well suited for clinical use. Moreover, with the advent of reprogramming technology, the development of methods to generate functional HSC from iPS cells promises to one day make it possible to generate unlimited numbers of perfectly matched, patient-specific HSC for therapy.
In the early days of HSC transplantation, the only graft source available was bone marrow, which was harvested from the pelvis or sternum under general anesthesia. While investigators were fine-tuning the details of conditioning regimens and graft composition, persuasive evidence emerged that a strong correlation existed between the absolute number of HSC present within bone marrow graft transplanted and the robustness of hematopoietic engraftment. Increasing the dose of HSC also resulted in significantly lower mortality from infectious complications post-transplant . This realization led to the search for means of increasing the number of HSC that could be harvested for transplant.
Studies in the 1970s–80s revealed that administering Pyran Copolymer to mice or dogs caused primitive colony-forming HSC to “mobilize” from the marrow into the circulation [105, 106]. Moreover, the dose and the number of times/frequency with which Pyran Copolymer was administered directly affected the resultant levels of circulating HSC, leading the authors to conclude that mobilized peripheral blood (mPB) represented a viable source of HSC for transplantation [105–107]. Subsequent studies showed that hematopoietic cytokines could mobilize primitive HSC with long-term repopulating ability into the peripheral blood with relatively high efficiency , and that immunoisolation could be used to obtain highly enriched primitive HSC from this cytokine-mobilized human peripheral blood, thereby removing malignant cells and making it possible to use autologous cells to treat hematologic malignancies [109–113]. The non-invasiveness and ease with which mPB could be collected (compared to bone marrow harvest), and the discovery that HSC were present at higher frequency in these mobilized products than in steady-state bone marrow quickly led multiple laboratories/centers around the world to begin replacing bone marrow with mPB in the clinical setting [114–119]. Indeed, the use of mPB as an HSC source has dramatically increased over the past 15 years or so, and now accounts for ~75% of HSC transplants from unrelated adult donors  and ~99% of autologous HSC transplants .
Interestingly, this dramatic shift has taken place without any hard clinical data to support the superiority of mPB, in terms of better patient outcome or safety , and despite concerns that the higher T-cell content of mPB might lead to higher GVHD risk [115, 121–124]. Support for this shift has come from numerous large, multicenter, randomized trials performed to-date, which have collectively shown that engraftment/reconstitution is often faster and more robust with mPB than with marrow, and may be associated with a decreased relapse rate, but this appears to depend upon the patient population and malignancy in question [125, 126]. However, as originally feared, these studies have also revealed that the use of mPB is associated with a significantly higher rate of chronic (and perhaps acute) GVHD than marrow, and offers no significant improvement in survival [114, 115, 124, 127–137].
Nevertheless, it has recently been suggested  that specific characteristics of the patients may dictate which source of the HSC graft is best. For example, cancer patients who have never undergone cytotoxic chemotherapy, and may thus be at increased risk for graft failure/rejection, might benefit from the use of mPB, given its ability to mediate more robust engraftment [138–141]. Similarly, patients over the age of 50, and those with serious coexisting diseases, are now routinely conditioned with reduced-intensity regimens for transplants. Since these milder conditioning regimens provide less intensive immunosuppression, the more vigorous engraftment potential associated with mPB transplantation could potentially be advantageous. On the other hand, given its reduced propensity to induce GVHD, bone marrow may be better most other patients, particularly those who are immunosuppressed from prior chemotherapy, since this suppression should lower the risk of graft rejection. Interestingly, studies by Morton and colleagues have shown that bone marrow collected after treatment with growth factors, i.e., mobilized bone marrow, may possess the same enhanced engraftment potential as mPB, but still retain the lower propensity to induce GVHD normally seen with steady-state bone marrow . However, these exciting initial results will need to be rigorously tested in well-designed randomized clinical trials comparing this HSC source with unstimulated bone marrow or mobilized peripheral blood.
Although the concept of using umbilical cord blood (UCB) as a potential source of HSC for transplantation first appeared in the scientific literature in 1972 , a decade passed before Makio Ogawa reported  the presence within human UCB of a unique class of primitive hematopoietic colony-forming cells that, upon replating, had the ability to robustly generate secondary colonies, including multipotential colonies. This report prompted numerous groups to begin examining UCB as a potential source of HSC for transplantation, but it was the pivotal work of Hal Broxmeyer that ultimately made it possible for UCB to move from the laboratory to clinical practice [144–146].
The first HSC transplant using UCB was performed in 1988 in a 5-year old patient with severe aplastic anemia due to Fanconi Anemia (FA) . This first UCB transplant required a tremendous collaborative effort between Arleen Auerbach, at the Rockefeller University, who had developed a method for accurate prenatal diagnosis of Fanconi anemia (FA) [148, 149], Hal Broxmeyer, at Indiana University, who had systematically analyzed the number of hematopoietic cell progenitors in cord blood with the ultimate goal of reconstituting hematopoiesis in humans [144, 146], and Eliane Gluckman, from Paris (France), who had shown that the in vivo hypersensitivity of FA cells to alkylating agents made it possible to successfully use attenuated dose conditioning in FA patients .
The UCB was collected during the birth of the healthy (confirmed in utero) sibling, cryopreserved, transported from Indiana to Paris in liquid nitrogen, thawed with no further processing, and transplanted into the preconditioned patient. The first signs of engraftment appeared on day 22, and gave way to complete hematological reconstitution and donor chimerism. The patient never developed GVHD, and is currently (>20 years post UCB transplant) healthy with complete long-term hematological and immunological donor reconstitution [147, 151]. This first successful transplant began a whole new era in allogeneic HSC transplantation, demonstrating that: 1) a single cord contained enough HSC to reconstitute hematopoiesis; 2) UCB could safely be collected at birth; and 3) UCB HSC could be cryopreserved and thawed without negatively affecting their repopulating ability .
In the years following this first successful transplant, UCB has become one of the more commonly used sources of HSC for allogeneic transplantation. While much of the original interest in cord blood was due to the possibility that it could be routinely collected at birth and cryopreserved for autologous use later in life (if needed), if the need arose, cord blood was soon found to exhibit many characteristics that make it an ideal HSC source. The first of these is its ready availability. A global network of cord blood banks has been established worldwide that has enabled the collection, cryopreservation, and distribution of over 600,000 UCB units to-date . Because UCB is donated in advance and banked, all routine testing has been completed, eliminating the long delay that is inherent to the use of marrow as a graft.
In addition to its ready availability, the relative immaturity of the immune cells present in unrelated UCB are far less likely to cause GVHD than their counterparts in marrow or mPB. As a result, in marked contrast to marrow and mPB, perfect HLA-matching is not necessary between donor and recipient for a UCB transplant to be successful. As a result, the patient’s chance of finding a suitable donor is greatly increased. It is also important to note that UCB is far less likely than marrow or mPB to transmit common viruses like EBV and CMV, which can be lethal for transplant recipients.
However, along with all of these benefits come several shortcomings that have yet to be fully resolved. It is now apparent that the immaturity of the cells present within UCB represents a double-edged sword. While their immaturity reduces GVHD, it also significantly delays engraftment/repopulation, lengthening hospital stays, and placing the patient at risk of serious complications . A second problematic issue with UCB is the limited volume that can be collected and the resultant small absolute number of HSC. While a typical UCB unit has ample cells to repopulate a newborn/child, there are usually insufficient numbers of HSC to successfully repopulate larger/higher weight adolescents and adults [145, 154]. This has largely limited the use of UCB as a graft to pediatric patients.
Two approaches have been taken to address the limited cell numbers in UCB. The first is to transplant multiple UCB units into the same recipient to bring the total nucleated cell number up to the required threshold for engraftment. This approach has proven to be successful from the standpoint of achieving successful engraftment. However, studies have suggested that the use of multiple cord blood units may be associated with increased GVHD , and have revealed an interesting phenomenon: when multiple UCB units are transplanted, both contribute to initial engraftment/recovery, but it is the HSC from only a single UCB unit that dominate hematopoiesis and ultimately produce long-term durable hematopoietic engraftment, with the cells from the other UCB units being lost over time [156, 157].
The second approach investigators are pursuing to combat the limited number of HSC within a typical UCB unit is ex-vivo expansion/manipulation to enhance numbers and/or potency of the cells collected from UCB and their engrafting capability. A number of new procedures have been attempted to achieve this goal, including various cytokine cocktails [158–160], novel matrices such as nanofibers [161–163], copper chelating agents [164, 165], transcriptional activators and inhibitors [166–176], and feeder layers often consisting of MSC and/or endothelial cells [177–183]. While these methods have all demonstrated that it is possible to significantly expand committed progenitor cells in vitro, with the exception of studies employing a feeder layer [177–183] (which would obviously complicate translation to the clinical arena), the ability to expand true long-term repopulating HSC present within UCB remains an elusive target [184–186] that is the subject of intense investigation, with recent advances suggesting this goal could be achieved in the near future .
The recent discovery that somatic mammalian cells can be epigenetically reprogrammed to the hematopoietic lineage either by passage through a pluripotent state or directly [1–5], has raised the exciting possibility of generating patient-specific blood cells, and promises to achieve the ultimate Holy Grail in hematology — production of large numbers of healthy autologous HSC for transplantation. The ability to generate patient-specific HSC from non-hematopoietic cells would eliminate the risks associated with the use of allogeneic cells and simultaneously avoid the danger of relapse that is inherent to the use of autologous hematopoietic cells in patients with hematological malignancies.
Despite this great promise and intense efforts [188–195], however, studies thus far [194–207] have revealed that it is not a trivial matter to generate functional HSC from pluripotent cells such as ES and iPS cells. Although cells that appear phenotypically to be hematopoietic stem/progenitor cells can be generated, the efficiency of this process is extremely low, making it unlikely that the numbers of HSC that would be required to repopulate a patient could be generated with current technology. A bigger problem that has become apparent is that the hematopoietic cells generated by forced differentiation of human iPS and ES cells do not pass the “gold standard” test for HSC functionality; that is, they do not provide long-term multilineage engraftment upon transplantation into suitable xenogeneic recipients [189, 194, 206–208]. In addition, hematopoietic stem/progenitor cells generated via reprogramming (either via an iPS intermediate or through direct reprogramming of fibroblasts) often exhibit lineage skewing when they differentiate in vivo, likely due to incomplete/inappropriate epigenetic reprogramming [189, 194, 203, 206–208]. The finding that epigenetic memory of the starting cell is a key determinant of its suitability for reprogramming of the resultant iPS cells to a given cellular lineage [209–214], prompted recent studies to derive iPS cells from primitive bone marrow-derived hematopoietic cells. These studies have confirmed that it is, indeed, far easier/more efficient to generate early hematopoietic cells from iPS cells that were generated by reprogramming primitive bone marrow cells . However, since a major impetus for using iPS technology is that HSC are exceedingly rare in vivo, using these rare hematopoietic cells as the starting population to generate iPS cells seems somewhat circuitous and undermines, to some degree, the advantages of generating HSC from iPS cells.
Two recent publications [216, 217] are the only two studies to-date to report the successful generation of transplantable human HSC from iPS cells. To accomplish this seemingly impossible task, these investigators used a clever trick: they transplanted undifferentiated iPS cells into immunocompromised mice and allowed teratomas to form. Since teratomas contain cells of all three germ layers, they should conceivably contain cells of the hematopoietic lineage. Indeed, both studies showed that HSC were formed within the teratomas, and once isolated, from either the teratoma  or from the marrow of teratoma-bearing mice , these HSC repopulated the hematopoietic system of immunodeficient mice, proving their functionality. These studies are of clear importance, since they unequivocally demonstrate that functional HSC can be generated from human iPS cells, thus validating the promise that iPS technology can one day be used to generate patient-specific HSC. However, these studies also highlight the great deal of work that still needs to be done to understand the intrinsic and extrinsic signaling that occurs during hematopoietic development in vivo to efficiently generate functional HSC in vitro via reprogramming technology . Combining these shortcomings with concerns over their long-term safety [196–199], it appears that several key hurdles must be overcome for iPS-derived cells to become clinically viable as a source of HSC for transplantation [194–207, 218].
Two very recent studies, one from the group led by Derrick Rossi and the other from Shahin Rafii’s group have taken a major step towards generating clinically viable HSC via reprogramming [219, 220], by showing that it is possible to bypass the need for a pluripotent intermediate step, and simply induce mature cells to adopt the phenotype and functionality of HSC by expressing a carefully selected set of transcription factors. In the study by Rossi and colleagues, the investigators showed that mature murine hematopoietic cells can be coaxed to revert to a more primitive induced-HSC (iHSC) state, and that these iHSC possessed clonal multilineage differentiation potential, were able to reconstitute stem/progenitor compartments in vivo, and were able to engraft upon serial transplantation. In the other studies, Rafii and colleagues showed that combining the expression of a cocktail of transcription factors with subsequent propagation on a serum-free instructive vascular niche monolayer induces highly purified non-hemogenic human endothelial cells to give rise to hematopoietic colonies containing cells with colony-forming potential and the ability to serially engraft immune-deficient mice. These two studies are highly significant, since bypassing the need for a pluripotent intermediate has eliminated many of the risks inherent to ES cell and iPS cell-based approaches to reprogramming. Indeed, these exciting new findings suggest it may become possible, in the not-to-distant future, to use reprogramming to generate transplantable HSC for the clinic.
For reasons that are poorly understood, the transplantation and engraftment of even low levels of HSC can render the recipient tolerant to the HSC donor. This phenomenon is now being exploited to combat autoimmune diseases and to reduce/eliminate the immune response to allogeneic solid organ grafts.
The seminal observation by Owen , over 40 years ago, that a high frequency of identical blood types existed in twin pairs of cattle led him to hypothesize that this was the result of exchange of hematopoietic cells between twins in utero. Further ground-breaking studies by Billingham and colleagues [222, 223] showed that sustained exposure to foreign antigens during this period produced permanent immune tolerance; thus, Owen’s dizygotic twins were rendered immunologically tolerant to one another as a result of this naturally-occurring hematopoietic cell “transplant” [224–226].
Rodent studies performed by Denman and Morton in the late 1960s/early 1970s [227, 228] revealed that it was possible to transfer an autoimmune disease present in New Zealand Black (NZB) mice to strains of mice that do not normally develop this disease, simply by transplanting hematopoietic cells. Subsequent work proved that the converse was also true, namely that transplanting hematopoietic cells from normal donors could ameliorate autoimmune disease when transferred to affected NZB recipient mice , and thus set the stage for evaluating the possibility of using hematopoietic chimerism as a means of inducing donor-specific immune tolerance in organ/tissue transplantation and to restore normal immune balance to correct autoimmune diseases [20, 222, 230, 231].
Unfortunately, although several decades have passed since the first demonstration that hematopoietic chimerism can be used as a means of inducing immune tolerance, the standard of care for organ transplant recipients and for patients with severe autoimmune diseases is still lifelong pharmacologic immunosuppression . A major factor that has precluded the widespread use of HSC transplantation to re-establish tolerance to self-antigens (for autoimmunity) or to induce tolerance to alloantigens present in transplanted solid organs/tissues has been concern regarding the morbidity and mortality associated with allogeneic HSC transplantation, arising from the toxicity associated with the conditioning regimen and from induction of GVHD. As discussed previously, the development of “reduced-intensity transplants” greatly reduced the risks associated with the preconditioning aspects of HSC transplantation, and made it possible to begin performing HSC transplantation in an outpatient setting, and on patients whose old age or medical comorbidities had previously eliminated them as candidates for HSC transplant [233, 234].
With respect to the risks of GVHD, an unexpected conceptual breakthrough occurred when van Bekkum found that it was possible to cure preclinical animal models of multiple sclerosis (MS) and rheumatoid arthritis (RA) by transplanting autologous HSC . This remarkable discovery suggested it might be possible to treat autoimmune diseases by HSC transplantation without the risk of GVHD. However, subsequent studies revealed that only antigen-induced autoimmune diseases (such as those van Bekkum had employed) could be cured by autologous HSC transplantation; spontaneous autoimmune diseases, which are thought to be polyclonal stem cell diseases , cannot. Nevertheless, van Bekkum’s unexpected results strengthened the case for using HSC transplantation for autoimmunity, and led to the first clinical autologous HSC transplant for systemic lupus erythematosus (SLE) .
Given the severe comorbidities present in patients with autoimmunity and those requiring solid organ transplantation, the development of reduced intensity conditioning regimens and the possibility of using autologous HSC in some settings have truly opened the door to the use of HSC transplantation for clinical tolerance induction. Indeed, substantial small and large animal preclinical data [238–259], and a rapidly expanding body of clinical evidence [237, 242, 260–268], have collectively demonstrated that transplantation of autologous or allogeneic HSC can induce tolerance to allogeneic organs/tissues, and reset the host’s immune system, leading to elimination of autoreactive T cells and correction of autoimmunity. This approach has successfully induced tolerance to a diverse range of tissues, including kidney [242, 259, 265, 267–271], heart [245, 268], skin [245, 255], liver [272–275], trachea , and vascularized composite tissue allografts (CTA) [255, 257]. It has also successfully reversed autoimmune disorders including SLE [227, 231, 236, 237, 252, 260, 261, 276], Type I diabetes [239, 241, 244, 249, 250, 262], pemphigus vulgaris [276, 277], and experimental allergic encephalomyelitis (a preclinical model for multiple sclerosis) [235, 238, 247, 248, 260, 278]. However, results of clinical trials using HSC transplantation to treat “true” multiple sclerosis have been more mixed; some trials have reported very positive impact on clinical scores and disease progression [278–281], while others have found either no improvement at all, or perhaps even a worsening of neuroinflammation and demyelination following HSC transplant [282, 283].
While only some autoimmune diseases can be treated by autologous HSC transplantation, allogeneic HSC transplantation has no such limitations. By virtue of the so-called graft-versus-autoimmunity effect, allogeneic HSC transplantation could theoretically promise a cure for any autoimmune disease. Unfortunately, allogeneic HSC transplantation represents a double-edged sword, since with the graft-versus-autoimmunity effect, comes the inherent risk of GVHD, just as occurs with the graft-versus-tumor (GVT) effect, discussed earlier in Section 2. Because transplanting highly enriched/purified HSC grafts can eliminate GVHD, it stands to reason that combining the use of nonmyeloablative preconditioning with enriched/purified HSC grafts should theoretically allow the safe, widespread implementation of allogeneic HSC transplantation for immune tolerance induction. Early murine studies performed by Weissman’s group were some of the first to establish that highly purified allogeneic HSC could, like BM grafts, induce immune tolerance to a donor-matched solid organ (heart), while maintaining the ability to reject third-party grafts [284, 285]. These studies also revealed that the timing of the solid organ transplantation relative to the time of HSC infusion was rather flexible, since long-term graft acceptance was achieved regardless of whether the heart and HSC transplantations were performed nearly simultaneously  or the HSC were transplanted months before the introduction of a donor-matched heart . Subsequent studies, also by Weissman and colleagues, extended these findings to the realm of autoimmunity, showing that nonmyeloablative conditioning combined with transplantation of highly purified HSC corrected a murine model of SLE .
Looking collectively at the preceding studies, it is clear that a standardized, “one-size-fits-all” approach to using HSC transplantation to induce immune tolerance will likely not be possible. Instead, the physician will need to tailor the approach to address the precise aspects of the immune system that need to be targeted, based upon the natural pathophysiology of the disease to be treated and the specific mechanisms by which the organ to be transplanted is capable of presenting antigen. While this may seem an insurmountable task, the effort is warranted, given the tremendous impact such an advance would have on the healthcare system and on the quality of life of tens of thousands of patients.
A great deal of preclinical and clinical evidence over the last 10 years has shown that the transplantation of HSC can also exerts a therapeutic benefit in non-hematopoietic diseases of multiple organs, including the liver, heart, and brain. The mechanism whereby these cells exert their benefit is, however, still an area of intense debate. In this section, we focus on liver disease/failure as a paradigm to illustrate the tremendous potential HSC possess as regenerative medicine therapies, and will then briefly summarize some of the most promising studies that have evaluated the application of HSC to the treatment of diseases/injuries of other non-hematopoietic tissues, including the heart and brain.
Given the shortage of available donor organs for transplant, a great deal of time and resources are being devoted to identifying alternatives to organ transplantation to treat liver diseases. Even if a solution to the limited supply of available donor organs could be developed, finding a compatible liver and receiving a transplant is, sadly, just the beginning of a high-risk process that will continue for the remainder of the recipient’s life. The transplant itself can fail within the first days post-surgery, requiring the immediate performance of a new transplant. Even if the transplant itself is successful, further intervention may be required to address surgical complications like vessel narrowing, bleeding, or clot formation. Patients are then placed on lifelong immunosuppression, to prevent the life-threatening risk of rejection of the donor liver. This necessitates that these patients are kept under close medical surveillance for the remainder of their lives, since their immunosuppressed state places them at risk of succumbing to infections, and to developing cancers such as leukemias and lymphomas.
Using cell therapy to repopulate/repair the liver would be far less invasive than replacing an entire organ, significantly reducing morbidity and the cost to the patient. Moreover, in some cases, autologous cells could be employed, eliminating the need for costly and dangerous lifelong immunosuppression. Furthermore, a predetermined mixture of cells of specific lineages/differentiative stages could be transplanted to ensure that rapid, perhaps short-term, engraftment would be obtained to quickly supply the patient with the requisite hepatic function. More primitive stem/progenitor cells of the same lineage could also be transplanted, providing, ideally, lifelong correction of the patient. This ability to transplant multiple cells at varying stages of development/differentiation could ultimately circumvent one of the major inherent difficulties with using cellular therapy to treat liver disease, namely, the fact that a certain critical mass of functioning hepatocytes must be present to maintain the basic metabolic requirements for survival [286–289]. Thus, in contrast to HSC transplantation in which the recipient’s hematopoietic system can be ablated to create “space” for the infused donor cells to repopulate the system, procedures for replacing the mass of the liver through cellular therapy will need to be designed in such a way as to ensure that a certain minimal degree of hepatic function is maintained to keep the patient alive while repopulation with donor cells takes place.
Among the non-hepatic sources of cells that could potentially be used to regenerate/repopulate the liver, the tissue in the body that has received the greatest amount of attention is the bone marrow. This attention is the result of numerous studies that have provided surprising yet compelling evidence that there are likely several different cell types within the marrow that could be used for liver repair/regeneration. Among these cells types, the one that has been the best studied is the HSC. As discussed in detail in Section 3, HSC can also be isolated from UCB and from mPB, making them readily available in relatively large quantities, and making it possible to harvest them from the patient to be treated using relatively non-invasive procedures.
Over the last several years, HSC have received a great deal of attention in the field of liver regeneration due to pioneering studies by Petersen and colleagues  showing that HSC appear to have the unexpected ability to give rise to hepatocyte-like cells in vivo following transplantation in rats, and other studies that quickly followed showing that bone marrow-derived cells could completely repopulate the liver of FAH mice [291, 292], correcting their disease phenotype. These initial findings led to a flood of activity in the field using a variety of rodent model systems (genetic lesions or chemical/physical injuries) to rigorously test the hepatocytic potential of HSC from various sources [291, 293–309]. Thus far, results from these studies have been rather hard to interpret, since each group has used slightly different criteria for isolating HSC, and each injury/disease model seems to have its own unique characteristics, resulting in differing outcomes, even when the same or very similar HSC populations are transplanted. However, some general trends that emerge are that human HSC derived from UCB, as opposed to those from bone marrow or mPB, consistently generate significant levels of hepatocytes following transplantation.
Another issue that has become clear is that HSC (like native hepatocytes) engraft at much higher levels when transplanted into recipients whose endogenous hepatocytes are defective as a result of genetic lesion or due to treatment with agents that prevent host hepatocyte replication. In fact, it appears that high levels of donor HSC-derived hepatocytes only occur in these disease/injury models if the transplanted cells possess some sort of proliferative/survival advantage over the endogenous host cells. However, studies using the fetal sheep model [310–313] have shown that it is possible to take advantage of the inductive microenvironment present within the developing fetal liver, and achieve high levels (up to 20%) of donor (human)-derived functional hepatocyte-like cells following the transplantation of highly purified populations of human HSC, in the absence of injury/disease. Moreover, the use of this model system allowed the investigators to demonstrate that a direct correlation existed between hepatocyte activity and the phenotype of transplanted human HSC, the HSC dose administered, and the source of cells used on a cell-per-cell basis (bone marrow, UCB, mPB). Remarkably, the human hepatocyte-like cells generated in this model retained the functional properties of normal hepatocytes, constituted hepatic functional units with the presence of human endothelial and biliary duct cells, and secreted human albumin into the circulation of the recipients.
Further complicating interpretation of preclinical studies to-date investigating the hepatocytic potential of HSC is the finding that the mechanism whereby the transplanted HSC generate hepatocytes differs depending on the nature of the existing disease/injury. In some cases, such as the FAH model, it appears that the donor-derived hepatocytes are generated almost entirely through the process of cell fusion, in which the transplanted HSC fuse with the host’s endogenous hepatocytes, generating cells that morphologically and phenotypically appear to be hepatocytes, yet contain the DNA from both the donor HSC and the host hepatocytes [292, 307, 314]. In other models, including the non-injury fetal sheep model, the hepatocytes appear to be generated almost exclusively through what seems to be true reprogramming/trans-differentiation of the transplanted cells into hepatocytes without exchange of any genetic or cellular elements between the host and the engrafted human cells [293, 297, 306, 308, 315, 316]. Another important caveat is that the phenotype and purity of the cell population transplanted also plays a role in whether hepatocytes are generated and, if so, whether donor-host cell fusion occurs [317–320]. The possible involvement of fusion raises concerns over the clinical use of these cells, since one can imagine that fusion may produce a relatively unstable genome, which, in the presence of existing liver disease/damage, could potentially induce hepatic tumors.
As a result of these concerns over the risks that may accompany the possibility of cellular fusion and the uncertainty of the outcome, the clinical use of HSC for repair/regeneration within the liver is still in its infancy . Despite these issues, however, the large number of successes seen in animal models has prompted 10 clinical trials using autologous bone marrow-derived cells in patients with liver disease, and 6 clinical trials have used either G-CSF-mobilized mPB or direct G-CSF injections to mobilize endogenous HSC into the circulation in the hopes that they would migrate to the liver and mediate repair/regeneration. The first 6 of the bone marrow-derived trials utilized unsorted bone marrow mononuclear cells [322–327], a population that likely contained both HSC and mesenchymal stromal/stem cells (MSC), while the remaining 4 used CD34-selected cells [328–331]. Three of the unsorted mononuclear cell trials were, in essence, uncontrolled feasibility studies, in which the number of patients in each case was small and no control arm was included [322, 324, 327]. These trials included patients with a variety of liver diseases/disorder, including hepatitis-associated cirrhosis, alcoholic liver disease, drug-induced acute liver failure, primary sclerosing cholangitis, decompensated cirrhosis, and cryptogenic cirrhosis. The endpoints utilized to assess treatment efficacy included quality of life, aminotransferase levels, model for end-stage liver disease (MELD) scores, and Child-Pugh (CP) score. CP is a clinical metric that is based on measurements of 5 clinical parameters: bilirubin, serum albumin, INR (prothrombin time), the presence/absence of ascites, and the degree of hepatic encephalopathy and its responsiveness to medication. Results of all but one  of these studies were promising, with no severe adverse events, and patients exhibiting improvements in one or more of the parameters being measured. Moreover, these benefits were often long-term, lasting several months. The one trial in which no benefit was seen  was randomized and controlled, and the authors concluded that the lack of benefit was likely due to this specific disease (decompensated alcoholic cirrhosis) rendering the liver relatively resistant to the regenerative effects of the transplanted cells.
The four trials to-date employing CD34-selected cells [328–331] were all uncontrolled trials with fewer than 10 patients. Keeping in mind this proviso, these pilot studies reported improvement in serum bilirubin and albumin and CP score following CD34+ cell infusion via hepatic artery or portal vein. Long-term follow-up in one cohort of 5 patients  demonstrated that 4 of the 5 patients exhibited improved clinical parameters that persisted for about 1 year post-infusion. The results of this follow-up are very encouraging and argue that the beneficial effects of BM-derived cells on cirrhosis may be fairly prolonged. However, the benefits in these trials did not come without any risk of adverse events, as one patient died of sepsis, and another developed hepatorenal syndrome, leading to discontinuation of one particular trial . Similarly, in study by Mohamadnejad and colleagues , the infusion of CD34+ cells via the hepatic artery was deemed to be unsafe and ineffective, since one patient died as a result of liver failure secondary to type 1 hepatorenal syndrome, and only 1 of the remaining 3 patients showed evidence of marginal improvement in clinical parameters.
As was the case with the studies using bone marrow-derived HSC, two-thirds of studies to-date using G-CSF mobilized HSC or G-CSF injection to treat liver disease have also been small patient number, uncontrolled pilots [332–337]. Further complicating data analysis and comparison between these trials, some studies utilized autologous HSC derived from G-CSF mPB, some administered G-CSF directly to mobilize the patient’s endogenous HSC, and some trials combined these approaches by injecting G-CSF and administering mPB-derived HSC. Despite these issues, these trials have shown that the use of G-CSF or G-CSF mPB appears to be safe and well tolerated, and G-CSF mobilization also appears to exert a significant therapeutic effect, as all 6 trials reported significant improvements in CP scores and MELD scores, improved survival rates, and improved sequential organ failure assessment (SOFA) scores, with some trials following patients for over 1 year after treatment.
Collectively, these studies with unselected or CD34-selected cells and with G-CSF mobilized mPB-derived cells provide hope that HSC may prove to be a valuable resource for cell-based therapies for liver disease. However, the results of these studies must be interpreted with some trepidation, given the limited number of patients enrolled in each trial, the frequent lack of appropriate controls, and the occurrence of adverse events in at least two of these trials. Furthermore, since the cells in all of these trials were autologously derived, there was no way for the investigators to assess the actual engraftment, persistence, or differentiative potential of the transplanted cells, leaving one to speculate as to the mechanism responsible for the observed clinical improvements.
As many metabolic disorders affect the developing central nervous system (CNS), and there are currently no therapies that can promise correction following brain/spinal injuries, the finding that HSC transplantation could improve liver function following disease/injury led investigators to study the ability of marrow-derived cells to contribute to repair/regeneration within the CNS. Early studies by Eglitis & Mezey  set the stage by clearly demonstrating that marrow cells were able to differentiate into both microglia and astroglia within the brains of murine BMT recipients. The ability of marrow-derived cells to give rise to astroglia was quite surprising, as astroglia are derived from the neuroectoderm, and should thus be developmentally distinct from the marrow. Two subsequent studies [339, 340] extended these initial findings by showing that intravenously infused marrow cells migrated to the brains of adult mice and differentiated into large numbers of cells expressing neuronal markers. While subsequent in vitro studies provided contradictory evidence that enriched populations of murine marrow-derived HSC do not, in fact, possess the ability to form functional neuronal cells capable of firing an action potential , the finding that mesoderm-derived marrow cells possess the ability to adopt neural cell fates in vivo, prompted further preclinical investigation and, eventually, clinical trials in patients with stroke/ischemic brain injury [342–347]. In similarity to the clinical trials presented above for liver disease, the clinical trials thus far for stroke have utilized multiple cell populations from different sources, and some have included G-CSF treatment to mobilize the patient’s endogenous HSC. These studies have shown that G-CSF and HSC can safely be administered to patients with stroke, and that prior labeling of the HSC with either iron microbeads or 99-Technetium allows the in vivo tracking of these cells for at least 24 hours post-infusion. Semi-selective accumulation of HSC at the site of injury and a trend towards a reduction in MRI ischemic lesion volume were also observed in one trial , but there are too few trials/patients to conclude with certainty the degree of improvement HSC mediate in this clinical setting, nor the mechanism that is responsible for any observed benefit. In similarity to the prior section detailing the use of HSC to repair/regenerate the liver, preclinical animal studies utilizing HSC from umbilical cord blood have shown that these cells also have the ability to migrate with some degree of selectivity to sites of injury/damage within the brain, and mediate repair following ischemic injury [349–355]. However, clinical trials have not yet been undertaken to assess the benefit cord blood-derived HSC will produce in human patients with ischemic injury/stroke.
Myocardial infarction (MI) is another major clinical target for which bone marrow cells are being explored to induce repair/regeneration. A large number of preclinical studies have been performed to test the ability of bone marrow and UCB-derived cells to repair the injured myocardium following MI [356–372], which were all set into motion by the pioneering studies conducted almost 15 years ago by Don Orlic [373–375]. These preclinical studies have led to an array of clinical trials using either HSC or G-CSF treatment to mobilize HSC to the infarct site [376–397]. Indeed, MI is probably the clinical setting in which HSC, and other bone marrow-derived populations, have been the most extensively studied, likely due in no small part to the fact that heart disease is one of the leading causes of death worldwide. With one exception , these preclinical and clinical studies have collectively provided compelling evidence that infusing HSC, or administering G-CSF to mobilize the patient’s endogenous HSC, can each lead to a reduction in infarct size, reduced left ventricular remodeling, and long-lasting functional improvement. In the one study that made a direct comparison , combining G-CSF injections with an infusion of mPB cells was found to be even more effective than administering G-CSF alone. One interesting finding that has also emerged is that the levels of engraftment are too low to support the conclusion that the marked beneficial effects on cardiac function is due to reprogramming/trans-differentiation of the transplanted cells to generate new cardiomyocytes. Rather, it seems that the infused cells are likely mediating their effects by: 1) dampening existent inflammation; 2) stimulating angiogenesis to restore blood flow to the infarcted zone; 3) restoring balance to the pro-apoptotic cytokine milieu of the damaged myocardium; and 4) releasing paracrine factors to promote repair and/or activate endogenous cardiac stem/progenitor cell populations. However, just as was the case with the previously discussed clinical trials in liver and stroke, the interpretation of the clinical data to-date in these trials for MI is not particularly straightforward. The main issue contributing to this lack of clarity is the use of somewhat poorly defined cell populations in many of these studies. G-CSF was originally developed and used to mobilize HSC, but it has now been shown that G-CSF can also mobilize MSC and EPC , populations that can exert their own trophic/paracrine and pro-angiogenic effects, respectively. Similarly, although the CD34 antigen has traditionally been used to isolate HSC, it is also present on endothelial progenitor cells (EPC) . As such, selecting CD34+ cells from G-CSF mPB likely yields a population enriched for both HSC and EPC. Nonetheless, future studies can likely resolve the issue of whether the observed effects are due solely to the HSC, and the trials to-date leave no doubt that bone marrow-derived cells can exert a pronounced therapeutic benefit following MI.
HSC represent the quintessential stem cell , and are likely the most studied and best understood stem cell within the body. Since their discovery/identification, HSC have been the focus of intensive research, and have proven to be invaluable and life saving following inadvertent radiation exposure and in the treatment of hematologic malignancies. In addition to the bone marrow, HSC can also be isolated from umbilical cord blood and from cytokine-mobilized peripheral blood, making them readily available in relatively large quantities, and making relatively non-invasive collection possible. The ability of HSC to completely repopulate the entire hematopoietic system would already guarantee them a valuable place in regenerative medicine, but the finding that hematopoietic chimerism can induce immune tolerance to solid organs and correct autoimmune diseases such as SLE, multiple sclerosis, and diabetes further broadened HSC’s clinical utility. This utility has increased yet again in recent years, with the demonstration that these cells, through a variety of mechanisms, promote repair/regeneration of tissues as diverse as liver, heart, and brain (and others that were not discussed, due to space constraints). By reviewing the existing literature on preclinical studies in small and large animals, and comparing/contrasting the results obtained in these models to the outcome in human clinical trials to-date, it is clear that mice are absolutely essential and invaluable tools for understanding basic HSC biology and for developing novel therapeutic approaches to treat human disease. However, these collective data also highlight the critical point that, despite their great value, mice are not simply “little humans”, and promising results obtained in murine models do not always translate directly into therapeutic success in the clinical arena. As such, it is essential to identify and make use of an animal model system that most closely approximates the specific human physiological and/or disease processes when performing preclinical safety and efficacy studies.
In conclusion, it is hoped that this review has provided the reader with a glimpse into the remarkable potential HSC possess, and highlighted their value as therapeutics in regenerative medicine.
Supported by Grants R01-HL097623 and R21-HL117704 from National Institutes of Health (NIH), and Grant NNX13AB67G from National Aeronautics and Space Administration (NASA)
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.