There are several different technical approaches to the isolation of hematopoietic stem cells (HSCs) with long-term repopulating ability, but these have problems in terms of yield, complexity, or cell viability. Simpler strategies for HSC isolation are needed. We have enriched primitive hematopoietic progenitors from murine bone marrow of mice from different genetic backgrounds by lineage depletion followed by selection of cells with high aldehyde dehydrogenase activity using the Aldefluor reagent (BD Biosciences, Oxford, U.K.). Lin−ALDHbright cells comprised 26.8 ± 1.0% of the total Lin− population of C57BL6 mice, and 23.5 ± 1.0% of the Lin− population of BALB/c mice expressed certain cell-surface markers typical of primitive hematopoietic progenitors. In vitro hematopoietic progenitor function was substantially higher in the Lin−ALDHbright population compared with the Lin−ALDHlow cells. These cells have higher telomerase activity and the lowest percentage of cells in S phase. These data strongly suggest that progenitor enrichment from Lin−cells on the basis of ALDH is a valid method whose simplicity of application makes it advantageous over conventional separations.
Hematopoietic stem cells (HSCs) are multipotent cells that are normally resident in the bone marrow and are responsible for producing all of the adult hematopoietic lineages. They can be enriched from bone marrow and used for transplantation purposes, and the ease of accessibility of the bone marrow makes them an attractive model system in which to investigate the stem cell biology of renewing tissues. Most enrichment procedures for HSCs have involved the fractionation of cells based on the expression of cell-surface antigens [1–6] or functional characteristics of the cells such as low-intensity fluorescence of Rhodamine 123 or Hoechst 33342 staining [7, 8]. There are several problems associated with these techniques. Primarily, the isolated populations, although highly enriched, still contain significant numbers of more mature progenitor cells . Second, the expression of surface antigens has been shown to vary among species, genetic background, and stem cell source . Some of these techniques require excitation at 350 nm (UV range), which may be harmful to the target cells. These procedures can be technically difficult and do not readily lend themselves to clinical applications. It is desirable, therefore, to develop alternative methods by which stem cell isolation may be achieved.
The aldehyde dehydrogenases (ALDHs) are intracellular enzymes responsible for oxidizing aldehydes to carboxylic acids [10, 11]. Members of this class are central to the processing of ethanol and amines produced during the catabolism of catecholamines and the conversion of vitamin A to retinoic acid [10, 11]. Both the hematopoietic progenitor cells and intestinal crypt cells display high levels of cytosolic ALDH and are relatively resistant to cyclophosphamide . To make use of this feature for purification of HSCs, several groups have developed fluorescent substrates for these enzymes. The first of these was based on the oxidation of dansylaminoacetaldehyde to dansylglycine . Jones et al.  used this substrate to purify viable hematopoietic progenitors that were able to confer delayed but stable engraftment of myeloablated recipients and lacked characteristic cell-surface HSC markers. This substrate suffers from the disadvantage of UV excitation, which could be mutagenic to the cells, and the emission spectrum of this substrate overlaps with other fluorochromes and cannot readily be combined with other stem cell markers. In 1996, Storms et al.  devised a new system that uses a visible light excitable fluorophore (BODIPY-conjugated aminoacetaldehyde) that is metabolized by ALDH to a carboxylate ion retained within the cell, allowing cells with high levels of ALDH to be isolated by fluorescence-activated cell sorter (FACS) because of their high fluorescence. Using a two-step strategy that combines the depletion of cells that express mature lineage markers with high activity of ALDH, highly purified hematopoietic-repopulating cells were purified from human umbilical cord blood  using the BODIPY-based substrate named Aldefluor. This substrate had not been applied to murine models, and we wished therefore to investigate its application for the purification of murine hematopoietic progenitor from bone marrow. Using lineage depletion followed by Aldefluor staining, we were able to isolate a highly purified population of hematopoietic progenitors from murine bone marrow of both C57BL6 and BALB/c mice. Furthermore, we have shown that Lin−ALDHbright cells show cell-surface marker characteristics of primitive hematopoietic progenitors and are highly quiescent and enriched for hematopoietic progenitor activity.
Materials and Methods
Bone marrow was removed from 8- to 12-week-old C57Bl/ 6JOlaHsd and BALB/c mice and depleted using a mixture of phycoerythrin (PE)–conjugated monoclonal antibodies directed against differentiated blood cells (CD11b, CD5, CD8a, CD4, Gr-1, CD45R, Ter119; all from BD Biosciences, City Name, ST) and PE magnetic beads (Miltenyi Biotech, Surrey, U.K.). Lin− cells were stained with the Aldefluor reagent (BD Biosciences) according to the manufacturer's instructions. All of the flow cytometry analysis was performed using a Becton, Dickinson (San Jose, CA) FACS Calibur, and cell sorting was performed using Becton, Dickinson FACS Vantage.
Phenotypic analysis of Lin−ALDHbright cells was carried out using allophycocyanin-conjugated antibodies to c-kit, CD44, CD45, Sca-1, CD34, Flk-1, CD29, CD81, PE-conjugated antibodies to Flk-2, and CD13- and biotin-conjugated antibody to CD38, which was subsequently detected by streptavidin peridinin chlorophyll protein. Statistical significance was assessed by Student's t-test.
Cell cycle analysis was performed using the Cycle Test Plus DNA reagent kit (Becton, Dickinson) in accordance with manufacturer's instructions. The data were analyzed using ModFit 3 (Verity Software House).
Telomeric repeat amplification protocol (TRAP) was carried out using TRAPeze Elisa Telomerase Detection Kit (Intergen) following the manufacturer's instructions. For each sample, three polymerase chain reactions were performed using 0.1, 1, and 10 ng of protein extract.
Colony-forming units–granulocyte-erythroid-macrophage-megakaryocyte (CFU-GEMM) assays were performed according to manufacturer's instructions (Stem Cell Technologies; Vancouver, Canada). In brief, approximately 3 × 104 bone marrow cells and Lin−ALDHlow or 1,000 Lin−ALDHbright cells in a total volume of 0.3 ml were added to 3 ml of Methocult medium (M3434), yielding triplicate cultures of 1.1 ml each. After 10–12 days in culture, the plates were scored for CFU–granulocyte-macrophage, CFU-erythroid, and CFU-GEMM colonies under light microscope.
Purification of Lin−ALDHbright and Lin−ALDHlow Cells
To enrich for hematopoietic progenitors, we carried out a lineage depletion of committed cells using anti-PE magnetic beads (Miltenyi Biotech) and antibodies directed against lymphoid, myeloid, and erythroid cells. The N, N-diethyl-aminobenzaldehyde inhibitor sample was analyzed first using FL1 and side scatter suggested by the manufacturer, and the R1 gate was chosen to represent the background level of cytosolic ALDH activity (Lin−ALDHlow; Fig. 1A). Staining of Lin− cells with Aldefluor but without the inhibitor caused a shift in fluorescence, which allowed us to define the Lin−ALDHbright cells in gate R2 (Fig. 1B). The R2 region comprised 26.8 ± 1.0% of the Lin− population in C57BL6 mice and 0.058 ± 0.01% of the total bone marrow cells (data derived from seven independent experiments). Similarly, in BALB/c mice, the R2 region contained 23.5 ± 1.0% of the Lin− population and 0.055 ± 0.01% of the total population (data derived from three independent experiments; supplementary online Fig. 1B).
Characterization of Lin−ALDHbright and Lin−ALDHlow Cells
Phenotypic investigation of Lin−ALDHbright cells in C57BL6 mice was performed using cell-surface markers common to the hematopoietic lineage (CD45; Figs. 2A, 2B), hematopoietic stem cell markers (c-kit, CD34, Sca-1, Flk-2, CD38, CD44; Figs. 2C–2J, 2M–2P), endothelial progenitor marker, Flk-1 (Figs. 2K, 2L), and mesenchymal stem cell markers (CD13, CD81, CD29; Figs. 2Q–2V). The Lin−ALDHbright population was comprised entirely of hematopoietic cells and devoid of mesenchymal or endothelial progenitors, as shown by specific marker staining. In addition, the Lin−ALDHbright population contained a higher percentage of cells that expressed the HSC marker c-kit compared with Lin−ALDHlow cells (Figs. 2C, 2D) but lacked expression of Sca-1 (Figs. 2G, 2H). Lin−ALDHbright cells were mostly devoid of CD34 (Figs. 2I, 2J) and CD38 expression (Figs. 2M, 2N), as demonstrated previously for murine hematopoietic stem cells in newborn and juvenile mice [17–19]. Higuchi et al.  have reported that a CD38+ subpopulation of HSCs appears before the age of 5 weeks and expands during adolescence. To ensure that we did not miss the developmental time point at which CD38 expression is observed in murine HSCs, we repeated the CD38 phenotypic analysis of Lin−ALDHbright at weeks 12, 14, and 16 (data not shown). We noticed no change in CD38 expression in the Lin−ALDHbright population between 8–12 weeks of development and 16 weeks, suggesting no expansion in CD38+ progenitors in the Lin−ALDHbright during this interval. Because CD38 expression is noticeable upon activation of juvenile HSCs in mice, it is tempting to speculate that the hematopoietic progenitors marked by high levels of ALDH activity are quiescent and might gain CD38 expression upon activation, an investigation that is now underway in our laboratory.
All Lin−ALDHbright cells expressed CD44 (Figs. 2O, 2P), a receptor for the cell-surface adhesion molecule hyaluronic acid, which has been shown to be synthesized by primitive hematopoietic cells and is involved in their lodgment to the bone marrow and their proliferation [20, 21]. The classical definition of HSC phenotype in mice includes low expression of Thy-1.1 and high expression of Sca-1; however, this is strain dependent and applicable only to mice expressing the Ly-6A/E and Thy-1.1 haplotypes [22, 23]. C57BL6/J mice express the Thy-1.2 haplotype, and although it has been shown that Thy1.2+ c-kit+ Sca-1+ Lin−/low cells are HSCs, the level of Thy1.2 expression is lower than Thy-1.1, causing Thy1.2+ cells to bleed into the negative gate. In view of these results, it was suggested recently that HSCs from Thy1.2 strains can be isolated by adding Flk-2 to the lineage cocktail . In addition, the loss of Thy-1.1 expression and gain of Flk-2 has been correlated to loss of self-renewal during HSC maturation, such that HSCs lacking Flk-2 have long-term repopulation ability, whereas HSCs with Flk-2 expression show short-term repopulating capacity . Our study showed that Lin−ALDHbright has fewer Flk-2+ cells compared with Lin−ALDHlow (Figs. 2E, 2F), suggesting that although separation on the basis of ALDH activity enriches for primitive progenitor cells, it is likely to include cells with both short- and long-term repopulating ability.
We repeated the same phenotypic analysis on BALB/c mice, which are of different genetic background to C57BL6 mice. We found that Lin−ALDHbright cells were comprised entirely of hematopoietic cells and mostly devoid of mesenchymal or endothelial cells (supplementary online Figs. 2A, 2B, 2M, 2N, 2T–2Y), as were Lin−ALDHbright cells in C57BL6 mice. Lin−ALDHbright cells showed expression of typical HSC markers, such as c-kit and CD44 (supplementary online Figs. 2C–2D, 2R–2S), and mostly lacked CD34 and CD38 expression (supplementary online Figs. 2K, 2L, 2P, 2Q). We did find very low expression of Sca-1 in both Lin−ALDHbright and Lin−ALDHlow cells (supplementary online Figs. 2I, 2J), and this is consistent with previous findings reported by Spangrude and Brooks  for the Lin− of BALB/c mice. We found a small percentage of Flk-2+ cells in the Lin−ALDHbright cells (supplementary online Figs. 2E, 2F), suggesting the presence of short-term repopulating cells in the Lin−ALDHbright cells. This analysis showed us that despite genetic background, Lin−ALDHbright cells show phenotypic similarities for the main HSC markers such as c-kit, CD34, CD38, and CD44.
Characterization of Lin−ALDHbright and Lin−ALDHlow Using In Vitro Colony Assays
Cell-surface staining predicted that the Lin−ALDHbright population is likely to be enriched for primitive hematopoietic progenitors. To investigate this possibility, we carried out in vitro colony-forming assays, which allowed us to estimate the plating efficiency of the cells and the nature of progenitors (Fig. 3, Table 1; supplementary online Figs. 3A, 3B). Comparison of the Lin−ALDHbright and Lin−ALDHlow populations showed that the Lin−ALDHbright contained all of the hematopoietic progenitor activity of Lin− cells in both C57BL6 and BABL/C mice, as demonstrated by colony assays (Fig. 3; supplementary online Fig. 3A) and an increased plating efficiency compared with Lin−ALDHlow and Lin− cells (Table 1; supplementary online Fig. 3B). Recent findings have also shown that the Lin−ALDHbright population of umbilical cord blood contains all of the hematopoietic progenitor activity and provides a better tool for resolving the functional differences in repopulating ability between human CD34+ cells . Despite the presence of c-kit expression, Lin−ALDHlow cells did not show any hematopoietic progenitor activity, which indicates that there are functional differences within the c-kit population that can be resolved on the basis of ALDH activity.
Cell Cycle Status of Lin−ALDHbright and Lin−ALDHlow Cells
Various studies have shown that murine HSCs enter the cell cycle less frequently [1, 8]. We performed cell cycle analysis on Lin−ALDHbright and Lin−ALDHlow cells of C57BL6 mice and found that the Lin−ALDHbright has the lowest content of cycling cells, averaging 10.8 ± 0.7% (Fig. 4A) compared with the Lin−ALDHlow population, which contained 41.5 ± 1.0% of cells in S phase (p = .002, n = 3). Similar results were obtained with the Lin−ALDHbright and Lin−ALDHlow cells of BALB/c mice (supplementary online Fig. 4A). This is consistent with the prediction that these cells contain more primitive progenitors that are quiescent or at least do not contribute to hematopoiesis at any given time .
Lin−ALDHbright Cells Have Higher Telomerase Activity Than Lin−ALDHlow Cells
Most cycling HSCs display telomerase activity, and this decreases as HSCs proliferate and differentiate into more mature cells that display negligible levels of this enzyme . To measure telomerase activity, we performed TRAP assays under linear conditions of product amplification using a range of dilutions of protein extracts that we have previously demonstrated to be a suitable range for cells with relatively high telomerase activity . We found that Lin−ALDHbright cells of both C57BL6 and BALB/c mice have significantly higher levels of telomerase compared with Lin−cells and Lin−ALDHlow (Fig. 4B; supplementary online Fig. 4B) and compared with murine embryonic stem cells that have been stably transfected with the reverse transcriptase unit of telomerase and thus show high levels of telomerase activity (Armstrong et al., unpublished data, 2004). The telomerase activity has been shown to correlate to the cell cycle status of the cells . It is therefore tempting to suggest that the cycling cells present in the Lin−ALDHbright population have much higher levels of this enzyme than the cycling cells in the Lin− and Lin−ALDHlow population.
HSCs are very rare cells that have the capacity to replenish the adult blood system throughout life. The ability to be partially expanded ex vivo and reconstitute the hematopoietic system of myeloablated recipients has made these cells very attractive for clinical applications and basic biology studies directed at reprogramming and differentiation of these cells. Because of their rarity, a lot of work has gone toward identification and enrichment of these cells. The most elegant methods that have resulted in purification of single cells with long-term reconstitution activity in murine models have used a two-step strategy that combines the depletion of mature blood cells using antibodies to cell-surface markers and positive selection using cell-surface markers highly enriched in HSCs such as Sca-1, c-kit, and Thy-1.1 [1, 2].
In this study, we have tried to identify murine hematopoietic progenitors on the basis of a general cellular function that HSCs possess in different species such as high level of ALDH activity [13–16]. Using dansyl ALDH as a substrate for aminoacetaldehyde, Jones et al. [13, 14] purified AA4.1−Lin−ALDH+ cells from murine bone marrow and showed that they were capable of delayed but durable engraftment in myeloablated recipients. Most important, the authors showed that AA4.1−Lin−ALDH+ cells did show low to undetectable levels of antigens presumed to mark HSCs, such c-kit, Thy-1.1, and Sca-1, and lacked spleen colony-forming activity. These phenotypic differences can be partially explained with the overlap in emission spectrum between the dansylaminoacetaldehyde substrate and other flurochromes. Given the advances made in identification of HSCs and mesenchymal stem cells during the past 8 years and synthesis of a new substrate for ALDH (Aldefluor) that does not require UV excitation, we felt that it was important to revisit the phenotypic characterization of murine progenitors isolated on basis of ALDH activity. Using the Aldefluor reagent, we were able to purify exclusively a highly potent hematopoietic progenitor population in both C57BL6 and BALB/c mice, which comprised 26.8 ± 1.0% and 23.5 ± 1.0% of the Lin− population, respectively, and was mostly devoid of mesenchymal or endothelial progenitors. At present, we cannot exclude the possibility that Lin−ALDHbright population may contain some nonhematopoietic stem cells circulating in the blood stream and finally attracted to the bone marrow. Although this is unlikely given that approximately 99% of Lin−ALDHbright cells are CD45+, more functional and phenotypic analysis should be conducted before any firm conclusions can be drawn.
The Lin−ALDHbright population contained all of the hematopoietic activity of Lin− cells, indicating the presence of highly potent hematopoietic progenitors. The cells purified with this method showed expression of certain HSC markers but not all. For example, 55% of Lin−ALDHbright cells in C57BL6 mice expressed c-kit, and 100% expressed CD44 while mostly lacking the expression of CD34 and CD38, as demonstrated for HSCs purified from newborn and juvenile mice [17–19]. However, Sca-1 expression was absent in the Lin−ALDHbright population of C57BL6 mice but present in the Lin−ALDHlow and Lin− cells at similar frequencies (4.3%; data not shown) to what has already been reported by others . This is consistent with previous observation of low-to-undetectable levels of Sca-1 in the Lin−ALDHbright population of C57BL6 mice [13, 14] but rather unexpected given the evidence from several publications describing the expression of Sca-1 in HSCs from C57BL6 mice. To confirm that this is not a technical artifact attributable to the antibody we had used for detection of Sca-1 (D7), we repeated the phenotypic analysis with a different antibody, E13 161-7 (data not shown), and found again lack of Sca-1 expression in the Lin−ALDHbright population of C57BL6 mice. This suggests that purification of hematopoietic progenitors on the basis of high ALDH activity marks a new progenitor-type population that has not been encountered before because of cell selection systems based on high expression of Sca-1.
One of the characteristics of stem cells is that they are largely quiescent. In support of this notion, Lin−ALDHbright population of both C57BL6 and BALB/c mice showed the lowest percentage of cells in S phase (11%–12.5%), although this seems to be higher than what has been previously reported for murine HSCs  and suggests the presence of more committed progenitor cells, which are likely to replicate more frequently. This possibility is supported by the presence of Flk-2+ cells in the Lin−ALDHbright population, which have been defined as short-term repopulating cells, as discussed earlier. This could also explain the higher percentage of cells assigned to the primitive progenitor population by this method from whole bone marrow (0.058% in C57BL6 mice and 0.055% in BALB/c mice compared with 0.01% for HSCs defined as Thy-1.1lowSca-1highLin−/lowMac1−CD4− by Morri-son and Weissman ) and the lower read out of progenitor activity (18 ± 2.4% compared with 83 ± 1% demonstrated for Thy-1.1lowSca-1highLin−/lowMac1−CD4−). Finally, this observation is consistent with in vivo repopulating studies, which have shown that Lin−ALDHbright cells provide a delayed engraftment in mouse and 10 times more Lin−ALDHbright human cells are needed to repopulate the nonobese diabetic/ severe combined immunodeficiency model compared with a more permissive model [14–16].
One of the problems associated with purification of HSCs from mice of diverse genetic backgrounds on basis of cell-surface markers is the difference in expression of these markers. For example, BALB/c mice mostly lack expression of Sca-1 in HSCs from bone marrow; instead they show expression of this marker in resting peripheral blood lymphocytes and activated T cells. We isolated Lin−ALDHbright cells from BALB/c mice and found that the cells purified by this method contained all of the hematopoietic activity of the Lin− cells, were highly quiescent, and showed high telo-merase activity, similar to the respective population in C57BL6 mice. This suggests that ALDH activity can be used to purify hematopoietic progenitors in mice of different genetic backgrounds. In addition, purification of progenitors on the basis of unique characteristics, such as the one described here, opens new avenues for investigating differences between HSCs from different strains of mice.
We would like to emphasize that marking of hematopoietic progenitors on the basis of ALDH activity using the new Aldefluor substrate provides an easy, reproducible, and convenient method for isolation of viable hematopoietic progenitors in clinical settings in which one aims to transplant long-term repopulating stem cells as well as transplant progenitor cells for short-term recovery after myeloablation therapy. In addition, it provides a powerful tool for isolating and comparing HSCs from different strains of mice for basic biology and aging studies. It will also prove useful for therapeutically directed studies that require isolation methods without the use of cell-surface antibodies for additional genetic manipulation such as nuclear transfer (in which the presence of antibodies may be harmful) from less terminally differentiated cells. Finally, it provides a powerful tool for isolating hematopoietic or other tissue-specific progenitors in cases in which their cell-surface phenotype is likely to be unknown such as from differentiation of embryonic stem cells.
The ability to obtain a population of cycling progenitor cells that can be used as targets for gene transfer and possession of relatively high levels of telomerase activity that facilitates the in vitro expansion of the cells before transplantation provides an excellent alternative to transduction strategies based on attempts of influencing the cell cycle of stem cells.
The authors would like to acknowledge Dr. Brian Shenton and Ian Harvey for help with FACS and Dr. Frans Nauwelaers for critical reading of the manuscript. We are grateful to Leukaemia Research Foundation for providing financial support to N.H. and Life Knowledge Park and BD Biosciences for financial support to L.L. I.D. is employed by BD Biosciences, whose product was studied in the present work.