Author contributions: O.P.-L.: study conception, collection and/or assembly of data, data analysis and interpretation, manuscript writing, final approval of manuscript; D.C.: flow cytometry experiment conception and design, collection and/or assembly of data, data analysis and interpretation, manuscript writing, final approval of manuscript; P.B.G.: NOD-SCID experiment conception and design, manuscript rewriting, final approval of manuscript; I.B.: data analysis and interpretation, manuscript writing; C.D.: assembly of data, statistical analyses; B.G.: bone marrow cell isolation; C.B.: provision of bone marrow sample; J.V.M.: participation in the research design, provision of bone marrow sample, and final approval of manuscript; M.P.: bone marrow sample provision, final approval of manuscript; A.B.-G.: discussion and manuscript revision; J.-J.L.: conception and design, data analysis and interpretation, manuscript writing, final approval of manuscript; M.-C.L.B.-K.: conception and design, data analysis and interpretation, manuscript writing, final approval of manuscript.
Disclosure of potential conflicts of interest is found at the end of this article.
First published online in STEM CELLS EXPRESS July 30, 2009.
Identification of prevalent specific markers is crucial to stem/progenitor cell purification. Determinants such as the surface antigens CD34 and CD38 are traditionally used to analyze and purify hematopoietic stem/progenitor cells (HSCs/HPCs). However, the variable expression of these membrane antigens poses some limitations to their use in HSC/HPC purification. Techniques based on drug/stain efflux through the ATP-binding cassette (ABC)G2 pump (side population [SP] phenotype) or on detection of aldehyde dehydrogenase (ALDH) activity have been independently developed and distinguish the SP and ALDHBright (ALDHBr) cell subsets for their phenotype and proliferative capability. In this study, we developed a multiparametric flow cytometric method associating both SP and ALDH activities on human lineage negative (Lin−) bone marrow cells and sorted different cell fractions according to their SP/ALDH activity level. We find that Lin−CD34+CD38Low/− cells are found throughout the spectrum of ALDH expression and are enriched especially in ALDHBr cells when associated with SP functionality (SP/ALDHBr fraction). Furthermore, the SP marker identified G0 cells in all ALDH fractions, allowing us to sort quiescent cells regardless of ALDH activity. Moreover, we show that, within the Lin−CD34+CD38−ALDHBr population, the SP marker identifies cells with higher primitive characteristics, in terms of stemness-related gene expression and in vitro and in vivo proliferative potential, than the Lin−CD34+ CD38−ALDHBr main population cells. In conclusion, our study shows that the coexpression of SP and ALDH markers refines the Lin−CD34+CD38− hematopoietic compartment and identifies an SP/ALDHBr cell subset enriched in quiescent primitive HSCs/HPCs. STEM CELLS 2009;27:2552–2562
Human hematopoietic stem cells (HSCs) are characterized by lifelong self-renewal capability, extensive proliferative capacity, and the ability to repopulate the bone marrow (BM) of immune-deficient mice (severe combined immunodeficient [SCID] repopulating ability). HSCs are usually defined by their primitive antigenic profile (Lin−CD34+CD38Low/− CD90+) and can be enriched based on these cell surface molecules. Sorting the CD34+CD38Low/− cell population has proven to be extremely useful for the characterization of human HSC and hematopoietic progenitor cell (HPC) cell compartments. However, the heterogeneity of membrane antigens  and their variable expression as well as the low potential of purified populations to repopulate BM have made HSC/HPC definition based solely on cell surface phenotype questionable. Recently, several groups have attempted to develop new purification strategies based on the unique biochemical or metabolic activity of these cells. Among the most promising strategies are two independent isolation protocols, one that uses the drug/stain efflux phenotype (SP) and the other that correlates the primitive hematopoietic cell compartment with high aldehyde dehydrogenase (ALDH) enzyme activity.
The first designated breast cancer resistance protein (BCRP1), ATP-binding cassette (ABC) G2, is the molecular determinant of the SP phenotype [2–4]. This membrane transporter is a member of the G subfamily of ABC proteins. ABCG2 is highly conserved in the vertebrate phylum  and is found in a wide variety of human and mammalian tissues. It plays an active role in excluding a battery of low-molecular-weight cationic substrates, including metabolic products, toxins, drugs, and dyes, from passing through the cell membrane . Hoechst low/negative cells, corresponding to cells that massively exclude the stain, have been designated SP cells by virtue of their typical cytometric profile in Hoechst red versus Hoechst blue [3, 7]. Therefore, stem cells , including HSCs, could be identified and purified on the basis of Hoechst 33342 exclusion [9, 10]. SP cells with the lowest Hoechst fluorescent profile have been shown to exhibit the antigen phenotype and cell proliferative capability of primitive HSCs distinct from SP cells with the highest Hoechst fluorescent profile . Therefore, whereas the SP compartment corresponds to a functional entity, it remains heterogeneous in terms of developmental stage within the hematopoietic hierarchy. It was recently shown that ABCG2 is preferentially expressed by immature human HSCs/HPCs and is responsible for the SP phenotype. It plays a regulatory role in early human hematopoiesis, as suggested by the enhancing effect of ABCG2 overexpression on the size of the progenitor pool [2, 12].
ALDH is an intracellular detoxifying enzyme expressed in the liver and is responsible for oxidizing aldehydes to carboxylic acids. Several isoforms of ALDH have been identified, with ALDH1-A1 being the primary active isoform expressed in HSCs/HPCs . ALDH functionality confers cell resistance to alkylating agents such as cyclophosphamide by deactivation of its metabolites. ALDH is also required for the conversion of retinol to retinoic acids. Inhibition of ALDH enzyme, using the dimethylaminobenzaldehyde (DEAB) inhibitor, promotes HSC self-renewal via the reduction of retinoic acid activity, demonstrating that ALDH is a key regulator of HSC differentiation . The ALDEFLUOR (Stem Cell Technologies, Grenoble, France, http://www.stemcell.com) technique has recently been developed to quantify ALDH activity level in viable cells [15, 16]. Using ALDEFLUOR BODIPY (4,4-difluoro-5,7-dimethyl-4-bora-3a,4a-diaza-5-proprionic acid)-aminoacetaldehyde (BAAA), a fluorescent synthetic substrate of ALDH, it is possible to isolate HSCs/HPCs using fluorescence-activated cell sorting (FACS) [17–20]. Similar to the SP phenotype, the ALDH activity level is considered as a stem cell marker because it is expressed in stem cells from several tissues, including hematopoietic, keratocytic, mesenchymal, neuronal, and endothelial stem cells [21, 22]. Whereas the cell fraction showing the highest expression of ALDH (ALDHBr) contains the largest number of CD34+CD38Low/− HSCs/HPCs [18, 23], it also includes primitive CD34− HSCs/HPCs , suggesting that, similar to SP cells, the ALDHBr compartment is heterogeneous, at least phenotypically.
With regard to the heterogeneity of the SP and ALDH fractions, we have assessed a multiparameter flow cytometric assay that combines SP and ALDH functionalities to analyze and refine the Lin− HSC/HPC compartments. Here, we show that association of the two markers allows identification of a more homogeneous HSC/HPC subpopulation within Lin− BM cells. Our study establishes that primitive Lin−CD45+ CD34+CD38Low/− cells can be found all along the gradient of ALDH expression, with a higher proportion in the ALDHBr compartment, and are associated with SP functionality (SP/ALDHBr cells). In conclusion, our results establish that the coexistence of SP and ALDHBr activities refines the primitive Lin−CD34+CD38Low/− HSC/HPC compartment. Indeed, the SP marker allowed for the discrimination of cells with different functional characteristics, in terms of quiescence and gene expression profile as well as in vitro and in vivo potential, within the primitive CD34+CD38Low/−ALDHBr compartment.
Human BM Collection and Cell Isolation
Samples were collected from healthy individuals undergoing hip surgery with their informed consent (Percy and CHR Robert Ballanger hospitals, France). Low density mononuclear cells (MNCs) were isolated on a Ficoll Hypaque gradient (1.077 g/ml density; Pharmacia, Orsay, France) and were incubated overnight on plastic culture dishes (TPP; ATGC, Marne-la-Vallée, France)at 37°C in a 5% CO2 atmosphere in order to remove adherent cells.
Immunomagnetic Isolation of Lin− Cells
Lin− cells were recovered from nonadherent MNCs by an immunomagnetic microbead technique, according to the manufacturer's instructions (StemSep Human Progenitor Enrichment Kit; Stem Cell Technologies). Cells were labeled with a cocktail of biotin-coupled antibodies raised against lineage-specific antigens: CD2, CD3, CD11b, CD14, CD15, CD16, CD19, CD56, CD123, and CD235a. Following a 30-minute incubation with biotin-labeled primary antibodies at 4°C, unlabeled cells were separated on a depletion column using the magnetic-activated cell sorting technology (Miltenyi Biotech, Paris, France, http://www.miltenyibiotec.com). This method resulted in 95%–98% pure Lin− cells.
FACS Analysis and Cell Sorting
Hoechst staining was performed according to Goodell et al. [25, 26] and others [2, 12, 27, 28], with slight modifications. Briefly, 106 Lin− cells/ml were suspended in prewarmed Dulbecco's modified Eagle's medium (Gibco, Cergy-Pontoise, France, http://www.invitrogen.com) supplemented with 5% fetal calf serum and 10 mM HEPES. Hoechst 33342 (Invitrogen, Cergy Pontoise, France, http://www.invitrogen.com) was added to a final concentration of 5 μg/106 cells/ml and cells were incubated at 37°C for 90 minutes. To confirm Hoechst active efflux by means of the ABCG2 transporter, verapamil hydrochloride (50 μM) or fumitremorgin-C (25 μM) were added to control samples before incubation with the Hoechst stain. Cells were then placed on ice for 30 minutes, in order to block the ABCG2 active efflux pump, washed twice with ice-cold Ca-/Mg-free phosphate-buffered saline (PBS), resuspended in ALDEFLUOR buffer and stained with synthetic BAAA ALDH substrate (ALDEFLUOR assay; Stem Cell Technologies) as described by the manufacturer. Baseline fluorescence was established on control cells first treated with Hoechst and ABCG2 inhibitors and then with BAAA and 10 μM DEAB, a potent inhibitor of ALDH. After a 30-minute incubation at 37°C, cells were centrifuged in cold PBS and suspended at 107 cells/ml in cold PBS for costaining with fluorescence-conjugated monoclonal antibodies, including anti-CD38 and anti-CD41, phycoerythrin (PE)-conjugated, anti-CD34 allophycocyanin (APC) conjugated, or anti-CD45 APC-cyanin7 (APC-Cy7) conjugated. Dead cells were excluded after staining with 7-aminoactinomycin D (7-AAD) (1 μg/106 cells; Beckman Coulter, Villepinte, France, http://www.beckmancoulter.com), using the PE-Cy5 channel. Prior to FACS analysis, cells were washed and resuspended in cold PBS at 107 cells/ml. For each experiment, 105 cells were analyzed.
Flow cytometry was carried out using a Becton Dickinson FACSDiva flow cytometer (Becton Dickinson, Le Pont-de-Claix, France, http://www.bd.com) equipped with an Enterprise IIC water-cooled laser (488/350 nm; Innova Technology, Coherent, Santa Clara, CA, http://www.coherent.com) and a Spectra Physics helium-neon laser (633 nm; Spectra-Physics Lasers model 127, Newport, Mountain View, CA, http://www.newport.com). ALDH reagent was detected by excitation at 488 nm and emission at 525 nm. Using the SSC versus FITC profile, we determined three Side Scatterlow fractions named ALDHLow, ALDHMid, and ALDHBr according to their ALDH activity level. Cells were analyzed according to Hoechst blue versus Hoechst red florescence. A gate was placed to select Hoechst low SP cells. Hoechst 33342 was excited at 350 nm, and fluorescence was measured with a 450DF20 BP filter (Hoechst blue) and a 675DF20 BP optical filter (Hoechst red). A 610 SP dichroic mirror was used to separate the emission wavelengths. For phenotypic analysis, PE or Pyronin Y was excited at 488 nm and emission was detected at 575 nm; APC was excited at 633 nm and emission was detected at 670 nm.
The SP and MP fractions were gated on the ALDHLow, ALDHMid, and ALDHBr fractions, sorted, and imunophenotyped using antibody combinations. The MP fraction corresponded to the non-SP main population of cells that exhibited high Hoechst fluorescence. In some experiments, the SP and MP fractions were sorted according to their CD34+CD38−ALDHBr expression.
Cell Cycle Analysis
A combination of Hoechst 33342 and Pyronin Y was used to stain Lin− SP/ALDH cells for a DNA/RNA content estimation as previously described . Quiescent (G0) and cycling (G1 and S+G2/M) cells were separated based on Hoechst/Pyronin Y fluorescence. Controls were established on cells without any Hoechst/Pyronin Y staining. For that purpose, 5 μM of Pyronin Y (10 ng/ml; Fluka-Sigma Aldrich, Saint Quentin, France, http://www.sigmaaldrich.com) was added into ALDEFLUOR buffer during ALDH substrate incubation for 20 minutes at 37°C. After washing, cells were suspended in cold PBS, immunostained, and analyzed using a FACSDiva flow cytometer.
Molecular Profile Analysis of Sorted Lin− SP/ALDH Cell Subsets
Reverse transcription of total RNA from 5 × 103 cells from the Lin−CD34+CD38−ALDHBr MP and Lin−CD34+ CD38−ALDHBr SP was carried out using a multiscribe TaqMan RT kit (Applied Biosystems, Courtaboeuf, France, http://www.appliedbiosystems.com). Quantitative polymerase chain reaction (PCR) was performed using an intercalating fluorescent substrate (SYBR Green PCR; Qiagen, Courtaboeuf, France, http://www1.qiagen.com). Ten picomole of each specific primer and 2 μl of cDNA were mixed to a final PCR reaction volume of 20 μl in LightCycler capillaries (Roche, Neuilly-sur-Seine, France, http://www.roche-applied-science.com). Amplification cycles (n = 45) were performed on a LightCycler (annealing temperature, 60°C). Data were normalized on the RPL38 housekeeping gene by a relative quantification based on the 2ΔΔCT method (NCBI GeoDataSets, GSE8230). The specific primers are described in supporting information Table S1.
Long-Term Liquid Culture and Semisolid Colony-Forming Unit Assays
To determine whether the Lin−CD34+CD38−ALDHBr SP or Lin−CD34+CD38−ALDHBr MP fractions were able to amplify and to sustain long-term hematopoiesis, cultures were set up from the different sorted cell subsets using SynH culture medium (Abcys biologie, Paris, France, http://www.abcysonline.com) supplemented with recombinant human (rh) stem cell factor, rhFlt3 ligand, and rh thrombopoietin (10 ng/ml; AbCys s.a., Paris, France, http://www.abcysonline.com). Representative aliquots were taken at several time points during liquid culture from day 0 to day 40 and cells were counted. Cultures were also assessed for their content of clonogenic HPCs using a colony-forming-unit assay. For that purpose, at different times of liquid culture, 1,000 cells from each subset were plated in 1 ml methylcellulose medium supplemented with cytokines according to the manufacturer's instructions (Methocult GF+ H4435; Stem Cell Technologies) in 35-mm Petri dishes. Duplicate cultures were incubated at 37°C in a humidified atmosphere containing 5% CO2. Colonies were scored using an inverted microscope after 10-14 days in culture on the basis of morphologic criteria.
Assessment of Engraftment and Growth Potential in Nonobese Diabetic-SCID Mice
Nonobese diabetic (NOD)/LtSz-SCID/SCID mice were housed in the Institut André Lwoff (Villejuif, France). Mice were irradiated at 3.25 Gy and received a dose of 250 μg of anti-CD122 antibody to abrogate residual natural killer cell activity 24 hours prior to intrabone injection (in the left femur) of 1,000 Lin−CD34+CD38−ALDHBr, Lin−CD34+CD38− ALDHBr SP, or Lin−CD34+CD38−ALDHBr MP cells. Mice were bled or sacrificed and their BM (two femurs and two tibias), spleen, and thymus were analyzed by FACS for the presence of human CD45+ (clone J.33) hematopoietic cells. In the case of intrabone injection, the injected femur was analyzed separately from the three other bones. In the case of positivity for human CD45 cells, cells were analyzed according to their repartition of B lymphocytes (CD19, clone J3-119) and myeloid cells (CD13, clone Immu103.44). All antibodies were purchased to Beckman Coulter (Villepinte, France, http://www.beckmancoulter.com).
On the day of sacrifice, mouse BM cells were studied for their content of human SP cells: 5 × 105 BM cells of each individual mouse were incubated for 90 minutes at 37°C with Hoechst dye at 5 μg/ml. Cells were then labeled with 7-AAD for viability together with antihuman CD45-APC-Alexa750 (clone H130; eBioscience, San Diego, http://www.ebioscience.com) and anti-murine CD45-APC (clone 30-F11; BioLegend Inc., San Diego, http://www.biolegend.com) for human and mouse discrimination.
All experiments and procedures were performed in compliance with the French Ministry of Agriculture regulations for animal experimentation.
Data are expressed as the mean of absolute number values ± standard deviation (SD). The significance (p < .05 was considered significant) between each condition and its control was determined using: (a) Student's t-test for paired samples, (b) Mann-Whitney and Wilcoxon nonparametric tests for independent values when data did not respect a Gaussian distribution, and (c) Pearson's correlation test for association between paired samples. The molecular profile and statistical cluster hierarchy analysis were realized using Log10 ratio data with TMEV software version 2.0 .
Simultaneous Hoechst 33342/ALDH Detection Identifies a Lin−ALDHBr Population Highly Enriched in SP Cells
Viable (7-AADLow) Lin−SSCLow cells were gated according to ALDH intensity into three subpopulations: ALDHLow (R1), ALDHMid (R2), and ALDHBr (R3); the ALDHMid gate was placed in the extension of the SSCHigh gate (R4) (Fig. 1A). We showed that SP and ALDH cells could be detected in similar proportions whether the ALDH and SP detection methods were performed alone (Fig. 1A, 1B) or together (Fig. 1C, 1D). Furthermore, the inhibitory effects of verapamil/fumitremorgin-C and DEAB were equivalent when added alone or together, demonstrating that the ABCG2 and ALDH functionalities are independent (Fig. 1A–1D).
Figure 1E shows that the dual detection method allowed the identification of SSCLow subpopulations that differentially express the SP marker based on ALDH enzymatic activity. These populations were named: SP/ALDHLow, SP/ALDHMid, and SP/ALDHBr. Analyses documented that SP cells could be found in all three ALDH subpopulations. However, ALDHBr was the fraction most enriched in SP cells, containing a fourfold higher percentage than in the ALDHMid and ALDHLow fractions (21.55% ± 8.57%, 5.09% ± 3.59%, and 4.15% ± 2.24%, respectively; p < .0001, n = 10) (Fig. 1F, histogram bars). Moreover, we found that SP cells from the ALDHBr fraction represented 50% of the total BM SP cell compartment against 29.7% ± 4.5% and 17.8% ± 3.1% for the ALDHMid and ALDHLow fractions, respectively (p < .05 and p < .001, n = 11) (Fig. 1F inset, curve). Therefore, the coupled SP/ALDH technology allows the quantitative analysis and isolation of a Lin−ALDHBr fraction highly enriched in SP cells.
The Dual SP/ALDHBr Population Includes the Primitive Lin−CD34+CD38Low/−CD90+ Cell Subset
Five gates were drawn according to SSCLow ALDH activity (R1, R2, and R3) and Hoechst fluorescence levels (R4 and R5), as described above; R4 corresponded to SP cells with a dull or low Hoechst intensity level and R5 corresponded to the MP cells, with brighter Hoechst staining (Fig. 2A). We first quantified the percentage of CD34+ cells in the Lin−SSCLow SP and ALDH compartments and showed that 72.60% ± 16.18% and 91.2% ± 13.1% of SP and ALDHBr cells were CD34+, respectively, without any significant difference between the ALDH cell subsets (data not shown). We further analyzed the coexpression of CD34 and CD38 antigens in the different gated subpopulations. Our results showed that the SP and ALDHBr subsets were enriched in CD34+CD38Low/− cells, with greater heterogeneity in the SP compartment (64.84% ± 17.64% versus 85.45% ± 6.39%) (Fig. 2A), in contrast with the ALDHMid, ALDHLow, and MP fractions (20.81% ± 9.58%, 4.94% ± 3.15%, and 11.34% ± 5.74%, respectively; p < .001, n = 8) (Fig. 2A). Interestingly, the coupled technology allowed the identification of SP/ALDHBr and MP/ALDHBr subsets highly enriched in CD34+CD38Low/− cells, as compared with other SP/ALDH and MP/ALDH fractions (95.59% ± 4.78%, 69.60% ± 17.78%, and 32.37% ± 26.24%, respectively; p < .001, n = 8) (Fig. 2B).
When cells were gated on SP and on Lin−CD34+CD38Low/− (R4 + R6) cells (Fig. 2C), we demonstrated that this compartment was heterogeneous in terms of ALDH activity. The ALDHBr compartment contained a significantly higher proportion of Lin−CD34+CD38Low/− SP cells than the ALDHMid and ALDHLow compartments (69.97% ± 13.24%, 17.33% ± 9.96%, and 4.21% ± 1.62%, respectively; p < .001, n = 8) (Fig. 2C). Therefore, the coupled SP/ALDH method demonstrates that the ALDH activity gradient discriminates different cell subsets within the heterogeneous primitive CD34+CD38Low/− SP fraction.
Analysis of CD90 expression, another marker of cell immaturity, demonstrated that similar results were observed in the Lin−CD34+CD90+ SP fraction (data not shown). Altogether, these results suggest that coupled SP/ALDH detection identifies a primitive CD34+CD38Low/−CD90+ SP/ALDHBr cell subset within the Lin− compartment.
SP/ALDH Dual Detection Reveals a Pure Fraction of Quiescent/G0 Cells Within the Lin−SSCLow CD34+ALDHBr Population
Pyronin Y, an RNA intercalating drug , is associated with Hoechst labeling to discriminate quiescent G0 cells from those in G1 and in S+G2/M. The different phases of the cell cycle were defined on the blue Hoechst/Pyronin Y biparametric plot according to Pyronin Y level after gating on SSCLowLin− (R1) and on CD34+ALDHLow (R2), CD34+ALDHMid (R3), or CD34+ALDHBr (R4) cells (Fig. 3A). SP (black dots) and MP (yellow dots) cells were further gated on these different populations (Fig. 3B), indicating the cell cycle status of SP/ALDH subsets according to ALDH intensity level (R1 + R2, R1 + R3, R1 + R4) (Fig. 3C). We showed that whatever the ALDH activity level, CD34+ SP cells were mainly in G0, as compared with MP cells, which also contained cells in G1 (Fig. 3C, 3D). Interestingly, all subpopulations contained a small proportion of cells in S+G2/M (Fig. 3C, 3E); the SSCLowLin−CD34+ALDHBr population contained the lowest percentage of cycling cells (2.19% ± 0.90%), as compared with CD34+ALDHMid (6.47% ± 1.72%; p < .01, n = 4) and with CD34+ALDHLow (12.78% ± 0.93%; p < .0001, n = 4) cells (Fig. 3E). Figure 3D shows that whatever the ALDH level, >90% of CD34+ SP cells were in the G0 phase of the cell cycle (90.08% ± 4.28%, 91.93% ± 7.87%, and 97.75% ± 1.19% for SP/ALDHLow, SP/ALDHMid, and SP/ALDHBr, respectively). These percentages were significantly lower in CD34+ MP cells (71.63% ± 9.45%, 58.2% ± 9.98%, and 84.63% ± 13.09% for MP/ALDHLow, MP/ALDHMid, and MP/ALDHBr, respectively; .001 < p < .05, n = 4). This confirms that SP functionality is associated with quiescence and demonstrates that quiescent SP cells are mainly in G0. Interestingly, the mean intensity level of Pyronin Y was lower in the SP/ALDHBr fraction, especially when compared with the SP/ALDHLow fraction (0.88 ± 0.66 AU versus 30.05 ± 12.12 AU, respectively; p = .03, n = 4) (Fig. 3C).
In conclusion, nearly all CD34+ SP cells were in G0 whatever the ALDH activity level. However, all G0 cells do not have the SP phenotype because the MP/ALDHBr fraction also contained a significant proportion of G0 cells. Therefore, our results show that (a) SP functionality more effectively discriminates quiescence than ALDH activity and (b) the association of the two markers allows the selection of a nearly pure population of CD34+ cells in G0.
Because both the ALDHBr SP and ALDHBr MP subsets are highly enriched in primitive CD34+CD38− cells and are mainly in G0, it was of interest to investigate whether the SP marker could discriminate cells with functional properties different from MP cells within the Lin−CD34+CD38−ALDHBr compartment. Therefore, further functional studies were restricted to sorted Lin−CD34+CD38−ALDHBr SP versus MP cells.
Combined SP and ALDH Markers Identify Cell Subsets With Different Characteristics of Immaturity, With the SSCLow Lin−CD34+ CD38−ALDHBr SP Cell Population Being the Most Primitive
Gene Expression Profile
Figure 4 shows a plot summarizing the expression pattern of some genes related to stemness, such as Notch-1, Oct-4, Gata-2, and Runx1, and some related to nesting, such as CXCR4, in both MP and SP Lin−CD34+CD38−ALDHBr cells. We showed that the transcript expression level of all tested genes was greater (up to 3.1-fold) in the Lin−CD34+CD38−ALDHBr SP cells than in the Lin−CD34+CD38−ALDHBr MP ones (.0005 ≤ p ≤ .0171, 3 ≤ n ≤ 5).
In Vitro Proliferative Potential
We compared the proliferation and clonogenic capability of Lin−CD34+CD38−ALDHBr MP versus Lin−CD34+CD38−ALDHBr SP cells at different time intervals (from day 0 to day 20) in cytokine-supplemented liquid culture. Figure 5A shows that the amplification rate of the Lin−CD34+CD38−ALDHBr SP cells was significantly higher than that of the Lin−CD34+CD38−ALDHBr MP fraction after 20 days in culture (647 ± 93 and 177 ± 125, respectively; p < .005, n = 3). These differences remained after 30 days in culture and only the Lin−CD34+ CD38−ALDHBr SP fraction was still viable at day 40 of culture (data not shown). Therefore, our results support the fact that Lin−CD34+CD38−ALDHBr SP and Lin−CD34+ CD38−ALDHBr MP cells have different proliferative abilities, with the Lin−CD34+CD38−ALDHBr SP fraction being the most proliferative (3.6×). Moreover, we investigated CD34 antigen expression during culture and we found that the Lin−CD34+CD38−ALDHBr SP subset maintained a twofold higher percentage of CD34+ cells than the Lin−CD34+ CD38−ALDHBr MP subset at day 20 (13.06% ± 1.29% and 6.28% ± 4.52%, respectively; p < .05, n = 4) (data not shown). Taken together, these results indicate that the absolute number of CD34+ cells after 20 days in culture was around eightfold higher for Lin−CD34+CD38−ALDHBr SP cells than for Lin−CD34+CD38−ALDHBr MP cells.
We also determined the cloning efficiency of both subpopulations during liquid culture. At day 0, we observed that the content of clonogenic progenitors was in favor of the Lin−CD34+CD38−ALDHBr MP fraction, with a cloning efficiency of 14.76% ± 5% (Fig. 5B, left panel), whereas that of the Lin−CD34+CD38−ALDHBr SP fraction was 6.05% ± 3.12% (p < .05, n = 5). However, whereas the cloning efficiency during the course of the culture remained quite stable for Lin−CD34+CD38−ALDHBr MP cells, as compared with day 0 (11.93% ± 3.06%; n = 5) (Fig. 5B, right panel), that of the Lin−CD34+CD38−ALDHBr SP subset increased up to fourfold, reaching 23.55% ± 2.07% (n = 7) at day 15 of culture (p < .0001 for Lin−CD34+CD38−ALDHBr MP versus Lin−CD34+CD38−ALDHBr SP cells). Taking into account the amplification rate and the cloning efficiency of both cell populations, which were, respectively, 3.65-fold and twofold higher in Lin−CD34+CD38−ALDHBr SP cells than in Lin−CD34+CD38−ALDHBr MP cells, the number of clonogenic progenitors was seven times more in the Lin−CD34+ CD38−ALDHBr SP fraction than in the Lin−CD34+ CD38−ALDHBr MP fraction (data not shown).
NOD-SCID Repopulating Ability
Total Lin−CD34+ CD38−ALDHBr, Lin−CD34+CD38−ALDHBr MP, and Lin−CD34+CD38−ALDHBr SP cells were injected into sublethally irradiated NOD-SCID mice (103 cells per mouse) using direct intrabone delivery. Their short-term (ST) and long-term (LT) SCID-repopulating cell content was evaluated 6 weeks and 12 weeks, respectively, after grafting. As shown in Figure 6A, the subpopulations allowed the engraftment of five of five (Lin−CD34+CD38−ALDHBr total cells), five of six (Lin−CD34+CD38−ALDHBr MP cells), and six of six (Lin−CD34+CD38−ALDHBr SP cells) mice at 6 and 12 weeks, without any significant difference in the level of engraftment, determined by the percentage of huCD45+ cells. Analysis of huCD19 and huCD13 expression revealed that injected cells were able to produce both B lymphoid and myeloid cells at 6 and 12 weeks (data not shown). The spleen did not show any engraftment, whatever the cell fraction injected (data not shown). We then investigated the ability of Lin−CD34+CD38−ALDHBr MP cells and Lin−CD34+ CD38−ALDHBr SP cells to maintain/generate SP cells in mice 12 weeks post-transplantation. For this goal, the BM of each mouse was labeled with Hoechst 33342 dye and with specific human and mouse antibodies. Figure 6B shows that 12 weeks after injection of human Lin−CD34+ CD38− ALDHBr SP cells, the murine BM contained viable (7-AAD−) human (huCD45+) SP cells. In contrast, despite the fact that Lin−CD34+CD38−ALDHBr MP cells were also able to generate a few SP cells, their proportion remained lower in most cases. When focusing on the entire cohort of mice, we showed that the proportion of SP cells generated by Lin−CD34+CD38−ALDHBr SP cells was five times higher than that generated by Lin−CD34+CD38−ALDHBr MP cells (0.7% ± 0.54% versus 0.13% ± 0.07% respectively; p < .01; n = 5 mice per group) (Fig. 6C).
The physiology of stem cells in human BM has become a central problem because of their numerous unique, yet elusive, characteristics. Here, we attempted to address better purification of HSCs/HSPs and present a novel, emerging approach to decipher the HSC hierarchy, applying metabolic stem tracers combined with canonical cell-surface markers.
Human HSCs represent a small fraction of BM cells and are usually purified according to their CD34+CD38Low/− phenotype. However, there are several limitations to purifying HSCs/HPCs exclusively based on cell surface determinants, because they may vary according to developmental stage, stem cell source, cell cycle status, regenerative process, and experimental conditions [1, 32]. Several teams have, therefore, introduced novel properties, such as Hoechst exclusion or ALDH activity, to purify HSCs/HPCs. The heterogeneity of both the SP and ALDH fractions with respect to CD34+ expression and also their primitiveness incited Bonnet's group and ours to combine the SP and ALDH detection systems to isolate HSC/HPC subsets . Results from Pearce and Bonnet's study  suggest that, in contrast to mouse HSCs, identification of human HSCs correlates better with ALDH activity than with Hoechst exclusion.
Here, we demonstrate the advantage of combining SP and ALDH selection to identify and purify human SP/ALDHBr cells that exhibit a primitive phenotype, characterized by the canonical CD34+CD38Low/−CD90+ phenotype, quiescent/G0 status, a high proliferative potential in liquid culture, and capability of engrafting and generating human CD45+ SP cells in NOD-SCID mice. Thus, our data indicate that dual SP and ALDH technologies refine the Lin−CD34+CD38Low/− progenitor compartment and define a new HSC/HPC distribution within this compartment.
In human BM, SP cells represent a rare population corresponding to about 0.05% of MNCs . When the Hoechst 33342 efflux function was applied to Lin− cells, alone or in association with ALDH activity, the proportion of SP cells exceeded 2%–5% of Lin− cells. With regard to the proportion of Lin− cells in MNCs (about 1%–2%), this proportion was comparable with that described by others . Similarly, the percentage of ALDHBr cells was not significantly different when the ALDEFLUOR technique was applied to Lin− cells alone or in combination with Hoechst staining (8.20% ± 3.19% versus 9.80% ± 6.34%, respectively). This suggests that the combination of the SP and ALDH technologies modified neither the functionality nor the viability of SP or ALDH cells, nor did it introduce technical biases.
It is generally held that the SP compartment involves only a small fraction of CD34+ cells, in contrast to the ALDHBr population, which is mainly composed of CD34+ cells (10.67% ± 13.1% and 91.0% ± 2.9%, respectively) [23, 24]. In our experiments, the percentages of CD34+ cells in the Lin−SSCLow SP and Lin−SSCLow ALDHBr fractions were 72.6% ± 16.2% and 91.2% ± 13.1%, respectively. This important difference in CD34+ cell enrichment was likely a result of the lineage depletion step that we incorporated into our protocol prior to SP/ALDH selection and that is not generally performed by others [25, 27].
By using combined SP and ALDH technology, we demonstrated that a fraction of Lin− human BM cells coexpresses both SP and ALDH functionalities, suggesting an overlap between these two markers. Indeed, all ALDH fractions contained SP cells, but with at least a fourfold higher percentage of SP cells in the ALDHBr subset (21.55% ± 8.57%) than in Lin− cells (2%–5%). This proportion was three- to fivefold higher than that described by Pearce and Bonnet , who showed that 0.04% of ALDH+ BM MNCs were SP cells. However, considering the percentage of Lin− cells (1%–2%) in MNCs, our results are in agreement with their data. Most importantly, we demonstrated the advantage of performing combined SP and ALDH selection on Lin− cells rather than on total MNCs, significantly improving the purification efficiency of cells that coexpress SP and ALDH stem cell markers.
Our results indicate that both the SP and ALDH fractions contain CD34+CD38Low/− cells, whose percentage increases with ALDH activity level. The dual SP and ALDH detection system allowed the identification of two Lin− fractions (SP/ALDHMid and SP/ALDHBr) highly enriched in CD34+CD38Low/− cells. When gated on either the CD34+ CD38Low/− or CD34+CD90+ subpopulations, the combined selection revealed their heterogeneity with respect to SP and ALDH activities, with a much higher proportion of CD34+CD38Low/− SP cells in the ALDHBr compartment. Therefore, the SP/ALDH method allows the selection of primitive CD34+CD38Low/−ALDHBr SP cells that also express CD90 within the Lin− compartment. A small overlap between the SP and ALDH subsets was reported by Pearce and Bonnet , suggesting that the identification of human stem cells was better achieved via ALDH activity than via Hoechst exclusion. Our data demonstrate that the coupled SP/ALDH method allows for the identification of an overlap between SP and ALDH cells that mainly concerns the primitive CD34+CD38Low/−CD90+ cells, suggesting that this combined method could be used to isolate primitive human hematopoietic cells.
It was previously demonstrated that most SP cells are quiescent and in G0/G1 . This was confirmed by Arai et al. [34, 35], showing that murine SP cells were in G0. By using Hoechst/Pyronin Y double labeling on sorted SP cells, we showed that human SSCLowLin−CD34+ SP cells are also in G0. Our combined SP/ALDH technique does not identify cell subsets exhibiting differential CD34+ cell cycle status  because whatever the ALDH expression level, all SP cells were found in G0. However, the intensity of Pyronin Y staining was lower in SP/ALDHBr cells than in SP/ALDHLow cells, suggesting a lower RNA concentration in the SP/ALDHBr fraction and mirroring their lower cell cycle-related transcriptional activity. In our experimental design, ABC pumps were inhibited by verapamil ; therefore, the low Pyronin Y level could not result from efflux through these pumps, ascertaining the specificity of the double-stranded nucleic acid Pyronin Y staining .
Whereas nearly all SP CD34+ cells are in G0 whatever the ALDH activity level, all G0 cells do not have the SP phenotype because ALDHBr MP cells also contain a significant proportion of G0 cells. Because both the ALDHBr SP and MP fractions shared similar phenotypic characteristics, it was crucial to explore whether, within the Lin−CD34+ CD38−ALDHBr compartment, the SP marker identifies cells with functional properties different from those of MP cells. When restricted to sorted Lin−CD34+CD38−ALDHBr cells, the proliferative capacity of SP cells was significantly higher than that of MP cells in long-term culture at day 20 and even longer (day 40). The fact that, in contrast to MP cells, the clonogenic potential of SP cells increased during culture is evidence that SP cells contain a higher proportion of pre-colony-forming cell primitive cells. When injected into NOD-SCID mice, Lin−CD34+ CD38−ALDHBr SP cells exhibited a repopulating ability similar to that of Lin−CD34+ CD38−ALDHBr MP cells. Interestingly, Lin−CD34+CD38− ALDHBr SP cells demonstrated the capability to maintain an SP phenotype 3 months after engraftment, as compared with Lin−CD34+CD38−ALDHBr MP cells. The fact that the Lin−CD34+CD38−ALDHBr MP compartment was also capable of generating a few SP cells suggests that this compartment may contain some HSCs, the SP phenotype of which could be versatile/reversible, as suggested in the murine system by Morita et al. . However, the higher capability of the Lin−CD34+CD38−ALDHBr SP compartment to generate SP cells is in agreement with their enriched primitive cell content.
Taken together, our in vitro and in vivo results provide evidence the presence of a functional hierarchy within the Lin−CD34+CD38−ALDHBr compartment, in which SP functionality identifies a compartment enriched in primitive HSCs/HPCs. Our molecular data showing that SP cells express higher levels of transcripts for stemness- and nesting-related genes than MP cells strengthens this assumption.
Therefore, our study allows us to propose a model indicating the overlap between SP and ALDH functionalities in the Lin−SSCLow compartment (Fig. 7). As shown, the coupled SP and ALDH technology indicates a gradient of quiescent CD34+CD38Low/− SP cells within the ALDH fractions, with the ALDHBr SP cell subset being the most enriched in primitive HSCs/HPCs.
Hematopoiesis is a dynamic and strictly regulated process orchestrated by HSCs and their supporting microenvironment. Given the role of SP and ALDH functionalities in HSC quiescence/self-renewal [14, 38–40], reflecting two processes involved in early stem cell regulation, the isolation of cells expressing both activities is of great interest to better understand these processes. Furthermore, because SP and ALDH are involved in stem cell multidrug resistance , this coupled technology is of clinical significance, allowing the comparison, quantification, and manipulation of normal and leukemic stem cells (LSCs) . Actually, it is hypothesized that LSCs are quiescent and closely associated with a specific niche environment, from which they can be forced to escape to expose them to antileukemic drugs . Targeting ABCG2 and ALDH activities simultaneously with specific inhibitors, currently under development, could open new therapeutic avenues for the treatment of neoplasia [39, 43–45].
We are indebted to Pr. Y. Masse, Dr. C. Blondeau, and Dr. A. Touma (Hôpital Intercommunal, Aulnay-Sous-Bois, France) and to Médecin Général M. Joussemet (Jean Julliard Army Blood Transfusion Center, Clamart, France) for supplying human bone marrow samples. This work was supported by grants from DGA (Délégation Générale pour l'Armement) and NRB association (association Nouvelles Recherches Biomédicales). O. Pierre-Louis was supported by grants from Conseil Régional and Conseil Général de la Martinique and from the SFH and NRB associations.