Many researchers consider the expansion of human hematopoietic stem cells (HSCs) as the holy grail of hematopoiesis research. Dozens of articles have been published that supposedly report the in vitro expansion of human HSCs, and many of these studies used the nonobese diabetic/severe combined immunodeficiency (NOD/SCID) xenotransplantation model as a readout system. In their recent article, Ando and colleagues1 claim to have demonstrated the first “direct evidence for ex vivo expansion of human hematopoietic stem cells.” This is based on using lentiviral gene marking to supply distinct clonal markers that can be used to trace the progeny of individual cells and the assessment of multilineage engraftment of individual clones in NOD/SCID mice as recipients. There is no doubt that this paper presents direct evidence for self-renewal of individual SCID repopulating cell (SRC) clones in vitro; however, the interpretation that HSCs were expanded would only be valid based on the assumption that exclusive true HSCs are read out in the NOD/SCID xenotransplantation model.
HSCs form a distinct primitive, pluripotent cell population that have the unique ability to continuously and in the long term repopulate all blood lineages—including myeloid, lymphoid, erythroid, and megakaryocyte lineages—of an irradiated recipient after bone marrow transplantation, commonly called long-term repopulation ability. This long-term repopulation ability is not assessed in cell populations defined by different in vitro assays such as colony-forming cells (CFCs), long-term culture–initiating cells (LTC-IC), and cobblestone area–forming cells (CAFCs).
Although the NOD/SCID model assays “a cell population different from most CFCs and LTC-ICs”2 (p1329) and is thus regarded by some as an appropriate assessment of HSCs, it must be kept in mind that it still is a surrogate assay (reviewed in Coulombel3 ). There is sound evidence that cells more committed than HSCs also contribute to engraftment in these recipients, which does not appear to be all that surprising, considering the vast difference on the proliferative demand placed on cells in a human versus in a NOD/SCID mouse as a recipient of a transplant and the relatively short time to readout of typically 6 to 10 weeks.
It was shown relatively early after introduction of the NOD/SCID xenotransplantation model that the types of cells present in NOD/SCID mice that underwent xenotransplantation differ depending on the kind and purity of the human cell population transplanted and that mature human hematopoietic cells in the transplanted cell population complicate interpretation of stem/progenitor cell readouts in these chimeras.4 In addition, data on transduction efficiencies and seeding efficiencies support the assumption that SRCs and CAFCs represent overlapping stem cell populations.5,6
When we transplanted aliquots of the same transduced cell population into NOD/SCID mice and nonhuman primates, we demonstrated that NOD/SCID repopulating clones were able to contribute to short-term repopulation in primates. However, none of the NOD/SCID repopulating clones appeared to contribute to hematopoiesis at 6 months or later after transplantation.7 This, together with the observation by Glimm et al8 that none of the more than 50 individual clones tracked by distinct vector insertion sites that were detected in the first month after transplantation were active later, suggests that in primates (most likely human as well as nonhuman) short-term hematopoietic reconstituting cells are distinct from hematopoietic stem cells and that the former contribute to NOD/SCID mice repopulation.
Taken together, even though it was clearly demonstrated that these cells resulted in multilineage repopulation of NOD/SCID mice, it can thus not be formally excluded that part or even all of the expanded cells reported by Ando et al1 would have behaved as short-term hematopoietic reconstituting cells rather than true long-term repopulating hematopoietic stem cells if they had been used as a transplant for a human. The holy grail remains elusive.
Measuring human hematopoietic stem cells
We appreciate the comments by Horn and Kiem regarding the definition of hematopoietic stem cells (HSCs) and its relation to severe combined immunodeficient (SCID) repopulating cells (SRCs) in our paper.1 An HSC is defined as a cell possessing both lymphomyeloid differentiation ability and self-renewal ability. Recent studies in mice,2 monkeys,3 and humans4,5 have elucidated the hierarchy within this population, which comprises short-term (ST) and long-term (LT) HSCs. The self-renewal ability of human LT-HSCs is usually assessed by serial transplantation experiments in SRC assays, since the life span of mice is much shorter than that of humans.
In this study, we analyzed 74 cultured SRC clones and identified 20 clones in more than 2 recipient mice. Eleven of these 20 clones were further assessed for their self-renewal ability by secondary transplantation experiments, with 3 clones, 3-23, 3-26, and 3-36, reconstituting 2 recipients at the same time. Therefore, we can reasonably claim that ex vivo–expanded LT-HSCs were present in at least 3 of the 11 tested SRC clones. As discussed in the paper,1 we could not discount the possibility that the clones that were undetectable in the secondary recipients may have corresponded to lymphomyeloid short-term repopulating cells (STRC-MLs). The clinical significance of this study resides, however, in the discrepant findings between the clonal analysis and the limiting dilution analysis (LDA), which is used in most clinically approved protocols for ex vivo expansion of HSCs. Our culture conditions resulted in a 5-fold expansion of SRCs by LDA and only a 1.5-fold expansion by clonal assay. Therefore, even if SRC expansion was shown by LDA in a clinical protocol, it might not guarantee the real expansion of SRCs. Our study highlights the fact that clonal analysis is required to gain accurate results regarding expansion of HSCs.
Horn et al reported that SRCs represent STRCs but not LTRCs in nonhuman primates by retroviral gene marking.6 This represented an important trial to determine the reliability of SRCs as a model for human HSCs in preclinical studies. However, we consider their conclusion to be premature for the following 4 reasons. (1) The conclusion was derived from the analysis of only 2 clones from a baboon. This is a case report and the reproducibility is not warranted. (2) ST and LT clones are different populations, as ST clones support only early-phase hematopoiesis and LT clones support late-phase hematopoiesis after transplantation.3,4 Therefore, it is not surprising that most SRCs at 6 weeks are STRCs and thus undetected in a baboon at 6 months after transplantation. The authors should demonstrate the presence of SRC clones in the recipients of serial transplantation, but not SRCs at 6 weeks, in a baboon 6 months after transplantation. (3) Lentiviral vectors transduce LTRCs more efficiently than retroviral vectors.7 (4) The most serious flaw is that the culture condition used in the study was not demonstrated to expand SRCs. To confirm this, it is required to detect common clones in multiple recipients after aliquots of the same transduced cells were transplanted.
Nonetheless, we agree that SRC assays still provide a surrogate assay for HSCs and endorse the viewpoint that the method needs improvement. Previously, we succeeded in humanizing the hematopoietic microenvironment in nonobese diabetic (NOD)/SCID mice by transplanting human mesenchymal stem cells into bone marrow.8 We also recently demonstrated that human LT-HSC clones could produce T cells, B cells, and myeloid cells up to tertiary transplantation over 1.5 years.9 Our future aim is to obtain the “holy grail” in SRC populations through these studies.
Correspondence: Kiyoshi Ando, Department of Hematology, Tokai University, School of Medicine, Isehara, Kanagawa, 259-1193 Japan; e-mail: firstname.lastname@example.org