|Home | About | Journals | Submit | Contact Us | Français|
The mammalian blood system contains a multitude of distinct mature cell lineages adapted to serve diverse functional roles. Mutations that abrogate the development or function of one or more of these lineages can lead to profound adverse consequences, such as immunodeficiency, autoimmunity, or anemia. Replacement of hematopoietic stem cells (HSCs) that carry such mutations with HSCs from a healthy donor can reverse such disorders, but because the risks associated with the procedure are often more serious than the blood disorders themselves, bone marrow transplantation is generally not used to treat a number of relatively common inherited blood diseases. Aside from a number of other problems, risks associated with cytoreductive treatments that create "space" for donor HSCs, and the slow kinetics with which immune competence is restored following transplantation hamper progress. This review will focus on how recent studies using experimental model systems may direct future efforts to implement routine use of HSC transplantation to cure inherited blood disorders.
Lifelong blood homeostasis is primarily the responsibility of hematopoietic stem cells (HSCs), which can self-renew for life while retaining the capacity to differentiate into distinct mature lineages. The numbers of HSCs and the decisions to undergo self renewal or differentiation are generally thought to be regulated by specific HSC-supportive niches , which may be agents of clonal selection and/or direct regulators of cell fate through their actions; yet deleterious cell-intrinsic mutations in HSCs, as they are passed along to downstream progenitors and mature cells, can cause immunodeficiency, anemia, various types of autoimmunity, and increase the chance of leukemia and lymphoma. In these settings, the replacement of the mutant HSCs with genetically normal HSCs is an important clinical goal.
Fortunately, HSCs possess several unique properties that have allowed them, in the context of bone marrow transplantation, to become the only type of stem cell in routine clinical use. First, although most HSCs normally reside in the extravascular space in the bone marrow, they have the remarkable ability to home back to their specialized niches in the bone marrow after injection into venous circulation. This ability to re-circulate appears to be linked to the normal properties of HSCs, since they constitutively migrate between the bone marrow, blood, lymph, and extramedullary organs under physiological conditions [2, 3]. Thus, HSCs can be injected intravenously rather than orthotopically into the bone marrow of recipients, making their clinical use relatively straightforward. Second, by enhancing the circulatory properties of HSCs using pharmacological agents such as granulocyte colony stimulating factor (G-CSF) or AMD3100 [4, 5], physicians now routinely obtain HSCs from peripheral blood of healthy donors rather than from bone marrow, a more invasive procedure. Finally, HSCs express a unique combination of cell surface markers that enable their purification. Although rigorous purification of HSCs prior to clinical transplantation is only rarely performed , HSCs can be isolated to near homogeneity as evidenced by single cell transplants with murine HSCs and 10-cell xenotransplants using human HSCs [7, 8].
Nevertheless, a number of clinical barriers remain which prevent the routine use of bone marrow transplantation for the treatment of blood diseases, particularly inherited disorders. Some of the most serious issues are graft-versus host disease (GVHD), balancing the side effects of cytotoxic pre-conditioning regimens against the chance of graft failure, and delayed immune reconstitution following transplantation. Because several excellent recent reviews exist on GVHD [9, 10], this review will focus on the latter two issues and on potential solutions offered by recent experimental studies, particularly as they relate to non-malignant disorders.
The concept of the need for "space" for transplanted hematopoietic cells has been controversial. Early experiments by Micklem and colleagues suggested that free space or empty niches were a limiting factor to the productive engraftment of bone marrow cells into unconditioned recipients . Later studies by Brecher, Micklem, and colleagues however, concluded that space was not an important factor to donor bone marrow engraftment and that specialized sites to allow for the proliferation of donor cells likely did not exist . Similar subsequent studies by several other groups concluded that transplanted bone marrow (containing HSCs) could readily displace endogenous HSCs in unirradiated recipients [13–15]. These studies were particularly important because HSCs are the only cells capable of populating the entire blood system for life, and the replacement of genetically defective HSCs with normal HSCs in the absence of conditioning would represent a significant clinical advance.
Other studies, however, came to very different conclusions. Gambel and colleagues found that donor marrow cells failed to engraft in unconditioned recipients, but could be stably transplanted into mice that had been pretreated with a depleting antibody against the major histocompatibility complex I (MHC-I) , a molecule that is highly expressed on enriched hematopoietic progenitors . While the anti-MHC-I antibody could have facilitated engraftment at least in part through the ablation of the host immune response, studies by our group have found only very low levels of engraftment of purified donor HSCs in unconditioned mice even when the hosts are genetically incapable of rejecting the graft [18, 19]. Severe combined immunodeficient patients (SCID) represent unique clinical cases in which marrow transplants are often performed without pre-conditioning of the host, since they are incapable of rejecting the donor cells . When these patients are not conditioned prior to receiving the transplant, donor HSC chimerism is less than 1% .
The studies that concluded that empty HSC niche space is not a limitation to transplantation all used unfractionated bone marrow transplants to reach their conclusions. At late timepoints following transplantation, total donor chimerism was measured in various tissues to estimate HSC chimerism in these studies. However, we have found that total hematopoietic chimerism, and in certain cases even progenitor cell chimerism, correlates poorly with and overestimates bone marrow HSC chimerism [19, 22]. Peripheral blood neutrophil chimerism appears to correlate well with HSC chimerism following transplantation of purified HSCs into unconditioned immunodeficient hosts [18, 19], but not after unfractionated marrow transplantation into wild type mice (D.B., unpublished observations). The experimental and clinical studies that transplanted highly or partially purified HSCs into unconditioned immunodeficient recipients found little evidence for high levels of sustained neutrophil chimerism or endogenous bone marrow HSC displacement [18, 19, 21, 23]. Taken together, these studies suggest that in the absence of conditioning, donor HSC engraftment is limited by the occupancy of appropriate niches in both humans and mice.
For patients suffering from inherited blood disorders, cytoreductive preconditioning regimens are applied for two main purposes: to create space for the transplanted donor cells as described above, and to ablate the immune response of the host to prevent rejection of the graft. The determination of the appropriate types of drugs and doses to accomplish these goals is not at all trivial. A unique clinical setting that has shed some light on this issue is the treatment of SCID, in which, as mentioned above, patients are incapable of rejecting the graft . Thus, the analysis of these patients allows for separation of the immunosuppressive effects of conditioning regimens from the myeloablative effects. Prior to marrow transplantation, SCID patients are either left unconditioned or are treated with reduced-intensity or fully myeloablative conditioning regimens, depending on the specific clinical parameters and the institution performing the transplant . A recent study by Cavazzana-Calvo and colleagues demonstrated that both unconditioned and reduced-intensity-conditioned patients generally failed to show sustained donor neutrophil chimerism when analyzed at least 10 years after transplantation, implying that few donor HSCs had engrafted at the time of transplantation [23, 25]. Consistent with this, patients showed very low levels of new T cell production by 10 years after transplant [23, 26]. In contrast, the neutrophils of patients that received full myeloablative conditioning regimens were exclusively of donor origin and new T cell production was sustained. These data imply that reduced intensity conditioning regimens may be insufficient to create space for transplanted HSCs and that fully myeloablative regimens must be employed. Fully myeloablative approaches using irradiation or DNA alkylating agents, however, are associated with serious side effects due to their lack of cellular target specificity and the risk of oncogenic DNA damage . Thus, the risks associated with these regimens discourage the use of HSC transplantation for the treatment of disorders that are not immediately fatal.
Some exciting recent clinical advances have come through depleting or inhibitory monoclonal antibodies due to the specificity and the lack of DNA damage that accompanies their use. Thus, antibody-based approaches for host HSC removal prior to transplantation could be attractive alternatives to the regimens currently in use. Yoshihara and colleagues recently demonstrated that when coupled with 5-fluoruracil, in vivo administration of a specific antibody against c-mpl prior to donor HSC transplantation resulted in increased chimerism over recipients conditioned with 5-fluorouracil alone . Our group recently demonstrated that administration of an anti-c-kit antibody to immunodeficient mice prior to transplantation was sufficient to remove host HSCs and allow extremely high levels of donor HSC engraftment in the absence of any additional conditioning , although future studies will be required to translate these findings into immunocompetent animals. These types of more specific approaches may facilitate the creation of HSC niche space while avoiding the DNA damage and other lasting side effects associated with the current fully myeloablative approaches. Thus, coupling highly specific reagents that create HSC niche space with highly specific immunodepleting antibodies and/or immunosuppressive agents that prevent rejection might increase the safety of conditioning regimens without reducing efficacy.
Following allogeneic marrow transplantation, patients are highly susceptible to opportunistic infections due to the immunosuppressive effects of GVHD and the conditioning regimen . GVHD can be prevented by the exclusion of donor T cells from the graft , but the conditioning regimens will invariably lead to at least transient immunosuppression due to the elimination of mature lymphocytes and, in some treatments, HSCs and progenitors. Although donor HSCs can eventually repopulate all mature blood lineages for life, the kinetics with which they do so after transplantation leaves an early period of immunological deficiency. One strategy to enhance the kinetics with which mature blood cells are regenerated is through the co-transplantation of progenitor cells (alongside HSCs) that have fewer developmental steps to take than do HSCs before reaching maturity. Moreover, unlike transplanted allogeneic mature T cells, lymphoid progenitors transferred at developmental stages prior to antigen receptor rearrangement and negative selection cannot cause GVHD [31–34]. Below we summarize the markers and functional properties of myeloid and lymphoid progenitors that have shown promise when used in conjunction with HSCs for rapid immune reconstitution following cytotoxic conditioning. Although many other downstream progenitors and differentiation pathways have also been extensively characterized, in this review we will focus on the progenitors and lineages that have specifically been implicated in mediating protection against pathogens following cytoreductive treatments.
Common myeloid progenitors (CMPs) were originally identified as cells in the mouse bone marrow that express c-kit and CD34 and low levels of FcγRII/III, but lack expression of Sca-1 and a panel of markers associated with lineage commitment . This ckit+ IL-7Rα− population was found to generate mature cells of the myeloid, but not the lymphoid lineage . Recent studies found that the Sca-1− mouse CMP population was fairly heterogeneous, containing multiple progenitors restricted to either the megakaryocyte/erythroid or granulocyte/monocyte fates [36– 39], and that the true CMPs express high levels of the transcription factor Gata1 and express low but positive levels of Sca-1 . In humans, CMPs can be isolated as cells that lack expression of lineage commitment markers and CD45RA, but do express CD34, CD38, and low levels of IL-3R α . At least 15% of these cells have the clonal ability to generate both granulocyte/macrophage colonies and megakaryocyte/erythroid colonies. Moreover, these cells can rapidly differentiate into IL-3Rα-lacking megakaryocyte-erythroid-restricted or CD45RA+ granulocyte/macrophage-restricted progenitors.
In a mouse model of allogeneic bone marrow transplantation, Bitmansour and colleagues found that transplantation of CMPs alongside HSCs, but not of HSCs alone, led to the protection of mice against lethal Aspergillus and Pseudomonas infections induced at early timepoints after lethal irradiation [42, 43]. Importantly, the donor CMPs need not be MHC-matched to the host to mediate protection , and thus could be relatively straightforward to obtain for clinical purposes. These proof-of-principle experiments demonstrated the potential utility of progenitor cell transplantation to augment HSC reconstitution.
T cells are the master orchestrators of the adaptive immune response, mediating both humoral and cellular immunity to intracellular and extracellular pathogens. Conditioning regimens that circumvent graft rejection will necessarily impair host T-cell function. Therefore, rapid, host-tolerant, T cell reconstitution following bone-marrow transplantation is extremely desirable. T cells develop in the thymus throughout life where they become self-MHC restricted and self-tolerant; however, the thymus lacks long-term self-renewing progenitors and must be continually seeded by cells from the bone marrow that home to the thymus through blood circulation . The identity of the bone marrow progenitor immediately upstream of the earliest thymus-resident T cell progenitor has remained controversial. An early candidate arose with the identification of the murine common lymphoid progenitor (CLP), which retains the ability to generate the T, B, dendritic cell (DC) and natural killer (NK) cell lineages, but can no longer differentiate into myeloid cells . Murine CLPs lack expression of a panel of mature lineage markers and have the surface phenotype IL-7Rα+c-KitloSca-1lo.
Initial characterization indicated that mouse CLPs were able to generate significant numbers of splenic T cells in sublethally irradiated recipients two weeks after intravenous injection. In contrast, it took over three weeks before comparable T cell chimerism was seen following transplantation of an equivalent number of HSCs (c-Kithi lin−Sca-1+). This delay in T cell generation was mirrored in the thymus where generation of donor HSC-derived CD4+CD8+ (DP) thymocyte development also lagged by over a week relative to CLP-derived thymocytes . These data suggest that CLPs are a good candidate for rapid T cell reconstitution from a bone marrow progenitor. In support of this hypothesis, Arber and colleagues demonstrated that following irradiation, transplantation of CLPs alongside HSCs, but not HSCs alone, led to the rapid protection of mice against lethal doses of murine cytomegalovirus; both MHC-matched and mismatched CLPs provided protection . As an alternative strategy, Zakrzewski and colleagues demonstrated that T cell progenitors derived from HSCs could be expanded ex vivo using OP9-DL1 stromal cells . These cells, when transplanted, underwent proper negative selection and did not mediate GVHD, yet could maintain protection against Listeria monocytogenes infections.
Recently, the efficiency and relevance of CLPs as T progenitors have been called into question [48–53]. Allman and colleagues found that the most immature mouse thymocytes (ETPs) were phenotypically more similar to the upstream multipotent progenitors (MPPs) than to CLPs for the expression of a number of cell surface proteins . Furthermore, since phenotypic CLPs are not present in Ikaros−/− mice, which retain nearly normal numbers of thymic ETPs, the authors argued that CLPs may not be important thymic progenitors . These data were used to suggest that a cell developmentally more proximal to HSCs, such as MPPs, may be the true bone marrow thymic T cell progenitor.
There are important caveats to consider for each of these arguments before excluding CLPs as thymic progenitors. First, it has been demonstrated that the c-Kit and Sca-1 levels of CLPs can be rapidly altered when CLPs are cultured on the stromal line OP9-DL1 that supports thymic development . Therefore, the reported differences in surface phenotypes between cells in the bone marrow and those in the thymus cannot be used to draw precursor-progeny relationships. Indeed, Schwarz and colleagues demonstrated that when injected intrathymically, CLPs increased expression of c-Kit and gave rise to phenotypic ETPs , a result confirmed in our own studies . Second, examination of the kinetics of mouse thymocyte development between intrathymically injected ETPs, CLPs, and MPPs clearly showed a high degree of similarity in the kinetics of early development of DP thymocytes between ETPs and CLPs. However, MPPs lagged behind by a couple of days (Fig 4, ); thus, it is difficult to justify using these data to argue that ETPs resemble MPPs more closely than CLPs developmentally. Finally, Ikaros null mice lack phenotypic CLPs, but they also lack phenotypic MPPs, since they lack Flk2 expression. An Ikaros-EGFP knock-in reporter revealed the presence of would-be Flk2+ MPPs, though these cells were functionally impaired in their ability to generate lymphocytes . Thus, both MPPs and CLPs may be phenotypically and/or functionally deficient in Ikaros null mice, even though ETPs are present in Ikaros−/− thymuses. These findings do not argue against either MPPs or CLPs as a physiologic thymic progenitor under normal conditions, but may instead reflect flexibility in input populations that can contribute to thymopoiesis under duress. Several groups have argued that the larger absolute number of thymic progeny downstream of MPP injection demonstrates that MPPs rather than CLPs are the major contributor to thymopoiesis [49–51]; however, it should be noted that MPPs can give rise to CLPs, and this precursor-progeny relationship opens the possibility that MPPs promote thymopoiesis only through a CLP intermediate. Thus, the larger readout downstream from MPP may be reflective of their ability to give rise to multiple CLP. Consistent with this model, CLPs produce an earlier wave of thymopoiesis than do MPPs [46, 51, 56]. It should be noted that several groups altered the original method of CLP isolation by selecting for cells that express high levels of AA4.1/CD93 but lack Sca-1 expression . Recent studies from our lab have shown that this method preferentially selects for B lineage-committed progenitors rather than CLPs, and that the use of Flk2 expression in conjunction with the original markers is a far superior method for the isolation of highly purified CLPs . These Flk2+ CLP yield over 50% thymic chimerism three weeks after transfer into sublethally irradiated recipients, underscoring their potency as T cell precursors .
Very recently, two prominent papers have suggested that the majority of ETPs retain myeloid potential at a clonal level, thus concluding that a lymphoid-committed progenitor cannot give rise to a physiologically relevant number of ETPs [52, 53]. That ETP isolated in bulk yield low-level myelopoiesis has been previously reported, and is not contentious . This low-level myeloid potential could be due to a rare subpopulation of ETPs, and would not be inconsistent with the majority of thymopoiesis arising from CLPs. On the other hand, if nearly all ETPs retain myeloid potential, then they cannot arise from an irreversibly committed lymphoid progenitor, such as the CLP. Only a clonal assay can distinguish between these two possibilities. In the study by Wada et al., very low-level myeloid potential was detected in bulk ETP cells, consistent with a rare progenitor maintaining myeloid potential. Furthermore, a clonal in vitro assay for myeloid potential of ETPs revealed only a small percentage of wells with both myeloid and T lineage potential. Myeloid potential was determined by the presence of Mac−1+ cells; when further assayed for additional markers, nearly all of the Mac−1+ cells in these cultures co-expressed CD11c, a known marker of dendritic cells . No definitive assays were performed to rule out the possibility that the “myeloid” cells in the clonal assays were dendritic cells. Importantly, it has been previously demonstrated that CLP can produce dendritic cells . Regardless of the identity of the Mac−1+ cells, if only a minority of the ETP population possesses clonal T and myeloid potential, it remains possible that the majority of thymopoiesis arises from CLPs. In contrast to Wada et al., Bell and Bhandoola conclude that 87% of ETP have both T and myeloid potential using a slightly altered clonal assay. Again, myeloid potential was determined by the expression of Mac-1, but it was not stated whether cells in this clonal assay expressed CD11c . Therefore, the possibility remains that many of the wells had T and dendritic cell potential instead, which does not rule out CLP as major thymocyte progenitors. Additionally, ETP have clearly been shown to have very minor myeloid potential in comparison to MPP using bulk in vivo and in vitro assays, such as colony forming assays on methycellulose (for example [48, 61–63]). This discrepancy with the clonal assay obtained by Bell et al. further raises the questions as to whether the readout obtained was truly myeloid and whether it accurately represented ETP lineage potential in vivo. An important but omitted control in these clonal assays would have been the inclusion of CLPs, which have no detectable myeloid potential in vivo. Thus, the identity of the bone marrow progenitor that is responsible for the majority of thymopoiesis remains contentious and will require additional experiments to reach a conclusive result. However, the rapid reconstitution of T cells downstream of CLPs has been experimentally documented as noted above, making it a useful progenitor in a transplantation setting irrespective of the above controversy.
In humans, common lymphoid progenitors were first found within the bone marrow as cells that express CD34, CD38, CD45RA, and CD10, but lack expression of a number of lineage commitment markers , and studies by our group found that these cells also express IL-7Rα transcripts . These cells, when assayed in xenogeneic transplants or in limiting dilution cultures, gave rise to T cells, B cells, NK cells, and DCs but not to myeloid cells. Other studies found that in the cord blood, only the CD10+CD7+ population contains CLPs, while the CD10+CD7− population contains myeloerythroid potential . Later studies, however, found that the majority of the T cell and NK cell potential is contained within the CD10− CD7+ population in the cord blood, while CD10+CD7− cells are mainly B lineage-committed . A recent study has shown that the majority of CD34+CD10+ cells in cord blood as well as adult bone marrow express CD24, which is a marker of B lineage-committed progenitors, probably resolving these previous discrepancies. They also demonstrate that CD7 is not expressed on the CD34+CD10+ population in adult bone marrow. Importantly, the CD34+CD10+CD24− progenitors contained true CLP in that they could generate B, T, NK and DC, but possessed little myeloid potential. Furthermore, CD34+CD10+ cells with similar potentials could be isolated from both adult blood and thymus, demonstrating that a lymphoid-committed progenitor likely transits from the bone marrow to the blood to the thymus to give rise to thymopoiesis in humans . Given that a lymphoid-committed progenitor appears to seed the thymus in humans, it would be highly surprising if this extremely important developmental step were not conserved in mice.
The field of bone marrow transplantation has made remarkable strides since the first clinical attempts by Thomas and colleagues . Improved methods for HLA typing and maintenance of donor registries, advancements in T cell depletion methods to minimize GVHD, new broad-spectrum antimicrobial drugs, and better conditioning regimens have brought us to a point where ~50,000 bone marrow transplants are now performed annually . Nevertheless, there is still considerable room for improvement at all stages of the process as it is currently performed. In an ideal setting, highly specific conditioning drugs, such as antibody reagents, would be used to eliminate host HSCs and graft-reactive lymphocytes prior to transplantation. Transplanted progenitor cells would then mediate rapid immunity without GVHD prior to lasting donor HSC-derived multilineage reconstitution (Figure 1). Translational studies using model systems can point us in the right direction on how to achieve these goals and, consequently, the safer use of HSC transplantation and replacement for a broad range of clinical purposes.
This work was supported by National Institutes of Health grants 5K01DK078318 (D.B.) and 5R01CA086065 (I.L.W.) and a fellowship from the Leukemia and Lymphoma Society (L.I.R.E).
Conflict of interest: Affiliations that might be perceived to have biased this work are as follows: I.L.W. cofounded and consulted for Systemix, is a cofounder and director of Stem Cells, Inc., and cofounded and is a former director of Cellerant, Inc. The other authors have no conflicting financial interests.