Current hematopoietic stem cell transplantation protocols rely heavily upon CD34+ cells to estimate hematopoietic stem and progenitor cell (HSPC) yield. We and others previously reported CD133+ cells to represent a more primitive cell population than their CD34+ counterparts. However, both CD34+ and CD133+ cells still encompass cells at various stages of maturation, possibly impairing long-term marrow engraftment. Recent studies demonstrated that cells lacking CD34 and hematopoietic lineage markers have the potential of reconstituting long-term in vivo hematopoiesis. We report here an optimized, rapid negative-isolation method that depletes umbilical cord blood (UCB) mononucleated cells (MNC) from cells expressing hematopoietic markers (CD45, glycophorin-A, CD38, CD7, CD33, CD56, CD16, CD3, and CD2) and isolates a discrete lineage-negative (Lin−) cell population (0.10% ± 0.02% MNC, n = 12). This primitive Lin− cell population encompassed CD34+/− and CD133+/− HSPC and was also enriched for surface markers involved in HSPC migration, adhesion, and homing to the bone marrow (CD164, CD162, and CXCR4). Moreover, our depletion method resulted in Lin− cells being highly enriched for long-term culture-initiating cells when compared with both CD133+ cells and MNC. Furthermore, over 8 weeks in liquid culture stimulated by a cytokine cocktail optimized for HSPC expansion, TPOFLK (thrombopoietin 10 ng/ml, Flt3 ligand 50 ng/ml, c-Kit ligand 20 ng/ml) Lin− cells underwent slow proliferation but maintained/expanded more primitive HSPC than CD133+ cells. Therefore, our Lin− stem cell offers a promising alternative to current HSPC selection methods. Additionally, this work provides an optimized and well-characterized cell population for expansion of UCB for a wider therapeutic potential, including adult stem cell transplantation.
Since the 1980s, hematopoietic stem cell transplantation and clinical research have mainly revolved around the cell surface protein, CD34, used as a marker for positive selection of heterogeneous hematopoietic stem and progenitor cells (HSPC) [1–4]. Although the true role of the CD34 molecule continues to be debated, CD34+ HSPC have been functionally defined as capable of generating progenitor-derived clones in vitro and by their potential in reconstituting the lymphomyelopoietic system in myelocompromised hosts [3, 5]. CD133 was reported as an alternative marker for positive selection methods targeting more primitive HSPC enriched with CD34bright cells [6, 7]. However, we have recently reported that although CD133+ cells had interesting ex vivo expansion potential, they still encompassed cells at various stages of differentiation .
Several studies have demonstrated that cells that lack CD34 and hematopoietic lineage markers (Lin−/CD34−) could engraft immunocompromised animal hosts and sustain long-term in vivo hematopoiesis [9–11]. Controversies then rapidly appeared by other groups suggesting limited hematopoietic engraftment potential to Lin−CD34− cells when compared with Lin−CD34+ HSPC [12, 13]. These investigators then hypothesized that small numbers of contaminating CD34+ HSPC could account for CD34− cell engraftment to the marrow . This controversy has been partly reconciled due to papers indicating the reversibility of CD34 expression, which may vary in vivo according to variable engraftment requirements [14–16].
Much of the problem with Lin− cell populations described so far is that they are poorly characterized. Modifications in previously described isolation protocols may also account for the variability of engraftment in immunocompromised animal models . Due to the need to find an optimized population, we have developed a reproducible strategy for primitive cell harvest.
Materials and Methods
Umbilical Cord Blood Collection and Mononuclear Cell Isolation
Collection of umbilical cord blood (UCB) specimens was from full-term deliveries scheduled for elective cesarean sections following hospital ethical regulations. Blood samples were diluted 1:4 in phosphate-buffer saline (PBS, Sigma Aldrich; Poole, UK; http://www.sigmaaldrich.com) supplemented with a citrate-based anticoagulant (0.6% ACD-A, Baxter; Maurepas, France; http://www.baxter.com), and bovine serum albumin (pH = 7.4, 0.5% fraction V, Sigma Aldrich) and referenced as ACD-A buffer. Diluted UCB was carefully overlaid in a 1:4 ratio onto a research-grade Ficoll-Paque solution (d: 1.077 g/cm3, Pharmacia Biotech; Uppsala, Sweden; http://www.pnu.com) prior to centrifugation (400 g for 30 minutes at 22°C). After extraction of mononuclear cells (MNC), cells were washed twice in ACD-A buffer, pelleted (400 g for 10 minutes), and resuspended in ACD-A buffer. Cell aliquots were spared for cell viability/enumeration using trypan blue exclusion method (Sigma Aldrich).
CD133+ Cell Immunomagnetic Selection
CD133+ cells were obtained from MNC after immunomagnetic separation using the CD133 mini-MACS selection kit (Miltenyi Biotec; Bergisch Gladbach, Germany; http://www.miltenyibiotec.com): labeling volume 500 μl/108 cells in ACD-A buffer containing Fc receptor-blocking reagent (100 μl, 5-minute incubation at 4°C) before adding colloidal superparamagnetic Miltenyi-activated cell sorter (MACS) MicroBeads conjugated to monoclonal mouse anti-human AC133/1 antibody (100 μl IgG1 isotype, 25-minute incubation at 4°C). Cells were then spun (5 ml ACD-A buffer, 400 g for 10 minutes at 4°C), resuspended in 500 μl ACD-A buffer, and applied to a chilled MACS-positive selection column (MS+/RS+) on a magnet. The column was rinsed with cool ACD-A buffer (4 × 500 μl). After magnet removal, CD133+ cells were eluted with 1 ml of cold ACD-A buffer. The CD133+ cell fraction was reapplied to a new column prior to cell enumeration and viability assays.
Lin− Cell Immunomagnetic Isolation
UCB MNC were incubated for 20 minutes at 4°C in human gammaglobulins (2% in PBS, Sigma Aldrich), to block nonspecific Fc receptors. Cells were then incubated for 30 minutes at 4°C with mouse monoclonal anti-human CD45, CD33, and CD7 antibody (Autogen Bioclear; Calne, UK; http://www.autogen-bioclear.co.uk) and anti-glycophorin-A antibody (Dako; Ely, UK; http://www.dako.dk). After centrifugational washes in ACD-A buffer (400 g for 10 minutes at 4°C), cells were subsequently labeled with Dynabeads Human IgG4 monoclonal anti-pan mouse IgG (Dynal; Wiral, UK; http://www.dynal.no) for 30 minutes. The cell fraction was then applied to a Dynal magnetic particle concentrator. After isolation, Lin− cells were counted and assessed for viability.
Flow Cytometry and Immunophenotyping
Cells were incubated in human gammaglobulins (20 minutes at 4°C, 2% in PBS; Sigma Aldrich) to block nonspecific Fc receptors. Direct immunolabeling was performed with fluorescein isothiocyanate (FITC), phycoerythrin (PE), or peridinin chlorophyll protein (PERCP)-conjugated monoclonal mouse anti-human antibodies (30 minutes at 4°C): CD38-FITC, CD34-PE/PERCP, CD7-FITC, CD2-FITC, CD3-FITC (BD-Pharmingen; San Diego, CA; http://www.bdbiosciences.com/pharmingen), CD133-PE conjugated (Miltenyi Biotec), CD16-PE, CD33-PE, CD56-PE (Dako), CD45-FITC (Autogen Bioclear), and glycophorin-A (IBGRL; Bristol, UK; http://www.blooddonor.org.uk/ibgrl). Intracellular CD34 labeling used anti-CD34-PERCP antibody and a PBS-based 0.01% saponin solution (Sigma Aldrich) for cell membrane permeabilization as previously described .
Indirect labeling was performed with primary monoclonal mouse anti-human antibodies: anti-CD164, anti-CD162, and anti-CXCR4 (30 minutes at 4°C; Becton Dickinson; San Diego, CA; http://www.bd.com). Secondary antibody labeling (20 minutes at 4°C) included goat anti-mouse IgG3-FITC (Southern Biotechnologies; Birmingham, AL; http://www.southernbiotech.com/home.htm) for anti-CD164 and F(ab')2 rabbit anti-mouse IgG FITC (Dako) for other primary antibodies.
Following appropriate washing procedures, cells were fixed in paraformaldehyde (1%, BDH; Poole, Dorset, UK; http://www.bdh.com). Fluorescent events were acquired on a Becton Dickinson FACScan flow cytometer with CELLQuest software prior to analysis with WinMDI software.
Cytokine-Stimulated Liquid Culture
Cells were seeded in duplicate in 9-cm2 tissue culture slide flasks (Nalge Nunc; Rochester, NY; http://www.nalgenunc.com) at 25 × 104/ml MNC, 2.7 × 104/ml CD133+ cells, and 2.7 × 104/ml Lin− cells. The liquid culture system consisted of Iscove's modified Dulbecco's medium (IMDM; Life Technologies; Paisley, UK; http://www.lifetech.com) supplemented with 10% fetal calf serum (FCS; PAA Laboratories; Somerset, UK; http://www.paa.at) and gentamicin (50 μg/ml, Life Science Technologies; Boone, NC; http://www.life-sciencetech.com). Culture systems were supplemented with two different growth factor conditions, respectively named K36EG: 20 ng/ml c-Kit-ligand (c-KitL), 50 ng/ml interleukin (IL)-3, 20 ng/ml IL-6, 6 U/ml erythropoietin (EPO), and 10 ng/ml G-CSF, or TPOFLK: 10 ng/ml thrombopoietin, 50 ng/ml Flt3-ligand (FL), and 20 ng/ml c-KitL. Cells were cultured at 37°C in 5% CO2 in a humidified atmosphere and enumerated weekly. Aliquots were removed for clonogenic assays when appropriate. Fresh medium and cytokines were provided every week and cells were maintained at a concentration below or equal to their respective original concentration. This was achieved by weekly harvesting with a sterile glass Pasteur pipette and pelleting the cells with a 400-g 10-minute centrifugation. After enumeration and viability assays, the cells were then dispatched in fresh flask appropriately. IL-3, IL-6, and EPO were purchased from R&D Systems Ltd. (Adbingdon, UK; http://www.rndsystems.com). G-CSF and c-KitL were a kind gift from Amgen Inc. (Cambridge, UK; http://www.amgen.com).
Long-Term Culture with Murine Stromal Cell Line MS5
The MS5 murine stromal cell line (American Type Culture Collection; Manassas, VA; http://www.atcc.org) was previously reported to successfully support HSPC differentiation in long-term culture-initiating cell (LTC-IC) assays [19–20]. Cells were subsequently seeded (MNC at 20 × 104/ml; CD133+ and Lin− cells at 0.4 × 104/ml) in 2.5 ml of LTC medium made of IMDM supplemented with 10% FCS, 10% horse serum (Sigma Aldrich), 5 × 10−7/ml hydrocortisone succinate (Sigma Aldrich), and 50 μg/ml gentamicin (Life Technologies), and cultured on a 15-Gy preirradiated MS5 stromal layer at 33°C at 5% CO2 in a humidified atmosphere. At weekly intervals, half of the nonadherent cells were removed and enumerated, and the flask was replenished with fresh LTC medium. After 6 weeks, nonadherent cells were collected and adherent cells were detached by trypsinization, and both cell fractions were then plated in clonogenic assays.
Clonogenic Assays in Methylcellulose
Cells to be assessed were cultured in 200 μl IMDM, 10% FCS supplemented with K36EG cytokine mix (concentrations as above), and 800 μl Methocult solution (H4230, Stem Cell Technologies; London, UK; http://www.stemcell.com) for 14 days at 37°C at 5% CO2 in a humidified atmosphere, prior to colony-forming unit (CFU) scoring.
When applicable, results are expressed as mean ± standard error (SE) from experiments performed in at least duplicate. Statistical significance was calculated by the Student's t-test.
Cell Purification and Phenotypic Characterization
The mean purity of CD133+ immunomagnetically selected cells was 94.3% ± 0.9% and represented 0.4% ± 0.04% of the original UCB MNC (n = 19). CD133+ cells coexpressed other surface markers as follows: CD34 (93.8% ± 0.9%), CD38 (62.7% ± 3.7%), CD164 (63.8% ± 6.0%), CD162 (69.3% ± 2.6%), and CXCR4 (67.7% ± 1.4%).
Lin− HSPC were characterized as a discrete cell population representing 0.1% ± 0.02% (n = 11) of original UCB MNC. The Lin− cell subset was negative for a wide range of hematopoietic lineage markers, including CD45, glycophorin-A, CD38, CD7, CD33, CD56, CD16, CD3, and CD2. A proportion of Lin− HSPC nonetheless expressed markers: A) reflecting their immaturity status [21–23], such as CD133 (7.0% ± 0.8%), CD34 (14.4% ± 3.6%), intracellular CD34 (16.2% ± 0.6%), and CD164 (16.0% ± 4.1%), or B) that were involved in HSPC migration, adhesion, and homing to the bone marrow (BM) [24–27], CXCR4 (48.6% ± 4.0%), and CD162 (96.7% ± 2.0%) (Fig. 1). Our negative isolation protocol appeared to be highly reproducible and isolated a rare primitive Lin− cell population.
CD133+ Cells Demonstrated a High Proliferation Potential in Cytokine-Stimulated Liquid Culture
Over 8 weeks in liquid culture, CD133+ cells proliferated more rapidly and yielded a significantly higher viable total cell number-fold increase (FI) than Lin− HSPC under both K36EG (p < 0.01) and TPOFLK (p < 0.05) stimulations (Fig. 2). In both culture systems, MNC were baseline and exhausted by week 7 of culture. Interestingly, TPOFLK cytokine mix induced a significantly higher viable cell FI when growing CD133+ cells for 8 weeks in culture (TPOFLK FI: 77.10 ± 0.27 × 107 versus K36EG FI: 1.53 ± 0.60 × 107, p < 0.05, n = 4). A similar pattern was observed for Lin− HSPC with a higher proliferation potential under TPOFLK synergism (week 8 FI: 33.20 ± 1.75) when compared with K36EG-stimulated liquid culture in which they stopped growing by week 6 (p < 0.001, n = 4). At first analysis, when compared with Lin− cells, CD133+ had a better ability to produce large cell numbers in TPOFLK-stimulated liquid culture.
TPOFLK-Stimulated Liquid Culture of Lin− Cells Maintains a More Primitive Population of HSPC than CD133+ Cells
The TPOFLK-stimulated liquid culture expansion system consistently maintained a higher proportion of CD133+ and Lin− cell-derived primitive colony-forming cells (CFC) when compared with the effect K36EG cytokine mix on these cells (Fig. 3) (p < 0.05, n = 4). This confirmed the value of TPOFLK synergism when expanding primitive HSPC in liquid culture.
After immediate selection, CD133+ cells were enriched for more CFC/104-seeded cells (485 ± 111) when compared with Lin− cells (251 ± 41) CFC and MNC (12 ± 2) (p < 0.05, n = 5). However, in TPOFLK-stimulated liquid culture, Lin− cells consistently maintained a higher proportion of CFC at weeks 4, 6, and 8 than CD133+ cells and MNC (Fig. 3). This difference was particularly significant at 6 and 8 weeks (p < 0.05). Interestingly, in TPOFLK-stimulated liquid culture, the peak in CFC maintenance/expansion from Lin− cells was observed at 6 weeks of expansion (34 ± 8 CFC/104 cells), correlating with their slow proliferation pattern. Under the same conditions, the highest CFC frequency from CD133+ cells appeared earlier at week 4 and gradually faded by week 8 (Fig. 3).
After selection, both Lin− and CD133+ cells were predominantly enriched with CFC from the erythroid line (56% and 72% of total CFC, respectively). Over 8 weeks, TPOFLK-stimulated liquid cultures gradually favored CFU-granulocyte macrophage (GM) expansion/maintenance, with CFC-erythroid exhausting from week 4 for Lin− cells and week 6 for CD133+ cells. However, from week 2 to week 8 of TPOFLK-supplemented liquid culture, Lin− HSPC yielded a higher ratio of CFC-GM than CD133+ cells (74% versus 58%, respectively, of total scored CFC) (Fig. 4).
Lin− Cells Contain More LTC-IC than CD133+ Cells and MNC
After 6 weeks in stroma-supported LTC assay, Lin− cells produced significantly more CFC than CD133+ cells (fivefold) and MNC (3,346-fold) (p < 0.05, n = 3) as shown by Figure 5. Taken together, these data showed that Lin− cells were enriched for more LTC-IC than CD133+ cells and MNC.
Current clinical protocols for HSPC collection mostly rely on CD34+ cells to determine the adequacy of the harvest . Several studies have also reported the CD133+ cell population as containing more primitive cells of interest for HSPC harvest [8, 28–32]. Although positive immunomagnetic selection was reported not to alter HSPC functional properties , direct antibody ligation is well known to potentially mediate cell-adhesion-molecule activity, including the CD34 molecule [22, 34–35]. Hence, negative selection protocols offer a real advantage to positive selection as they do not cause activation of HSPC adhesion molecules.
Conflicting studies previously proposed the human hematopoietic stem cell phenotype to be Lin−CD34+ and/or Lin−CD34− . Few published methods fully characterized the separated Lin− cells in terms of frequency, phenotyping, or LTC-IC enrichment . Moreover, most protocols used numerous primary antibodies to achieve negative isolation and sometimes required further CD34+ separation through positive immunomagnetic or fluorescence-activated cell sorting [36–38]. The cost of such prolonged protocols previously reported in the literature largely prevents regular clinical application (certainly in the European health care systems).
Unlike most protocols described so far, our negative selection protocol requires only four primary antibodies to separate a highly primitive, pure Lin− cell population, lacking many cell membrane markers associated with lymphohematopoietic cell commitment. This study hence offers a relatively simple, rapid, cost-effective standardized negative immunomagnetic selection method to isolate primitive Lin− cells from UCB. Our work also demonstrated little interpatient sample variability in terms of the Lin− cell subset frequency (0.1% of MNC), which was four times lower than the CD133+ cell frequency (0.4% of MNC).
In this study, freshly isolated Lin− cells were a phenotypically primitive population, with subsets expressing early HSPC markers such as CD133, CD34, and CD164, while lacking CD38, which is commonly reported as an activation marker associated with HSPC maturation [39–40]. Moreover, a significant proportion of Lin− cells expressed intracellular stores of CD34 molecules, with such cells previously reported to identify highly primitive HSPC [8, 21, 41].
Furthermore, Lin− cells were highly positive for CXCR4 (>48%, the stromal-derived factor-1 receptor) and CD162 (>96%, also known as P-selectin glycoprotein ligand-1). Both molecules were previously reported as essential for HSPC homing and adhesion when repopulating/developing in the BM [24–27, 42]. High expression levels of these markers may also correlate with the circulatory nature of UCB cells. CD162 is also reported to be highly expressed by primitive BM CD34+CD38− HSPC . Interestingly, CD34, CD164, and CD162 are cell adhesion molecules belonging to the sialomucin family, which has been proposed to act as a negative regulator of hematopoiesis, perhaps correlating with the high mobility of HSPC [43–45].
By isolating UCB Lin− cells containing both primitive CD34+ (and CD133+), as well as promising immature CD34− cells (a proportion of which expresses the CD34 protein intracellularly), our depletion method satisfies both theories on the phenotype of engraftable HSPC to be CD34+ or CD34−. We and others have previously shown that CD34 can indeed be present intracellularly at the mRNA and/or the protein level without being yet localized to the cell surface [16, 21, 41, 46].
Of interest, we have shown that both CD34− and CD34+ cell subsets harvested with our methods had not yet fully committed to lymphoid or myeloid lineages, as they did not express the leukocyte-common antigen CD45 or the erythroid marker glycophorin-A . This emphasizes the primitive nature of this Lin− cell population. The Lin− cells described herein were separated from UCB and most likely represent a developmental intermediate stem cell source between fetal tissues and adult BM. This confers an interesting potential for Lin− cells in stem cell plasticity studies. Our Lin− cell population may indeed encompass multi-/totipotent cells with transdifferentiation potential, because recently we have been able to induce their cross-development into neuronal precursors .
The primitive phenotype of Lin− cells described herein correlates with their high LTC-IC content. Although LTC-IC assays are difficult to standardize and may lead to variable results , they still provide key information regarding cell subset primitivity. Several studies have inferred LTC-IC to be closely related to, if not overlapping with, transplantable repopulating stem cells . Our Lin− cell isolation protocol gave 3,346-fold LTC-IC enrichment from the original MNC population and a fivefold enrichment over the CD133+ cell separation. After isolation, Lin− cells showed significantly lower CFC potential but were significantly enriched with more LTC-IC than CD133+ HSPC. The inverse relationship between CFC versus LTC-IC content concerning freshly isolated CD133+ and Lin− cells may be explained by LTC-IC being more primitive than CFC .
To further comparatively assess Lin− cell and CD133+ cell functional characteristics, we investigated both their proliferation ability and their clonogenic potential in cytokine-stimulated liquid culture. Similar to the LTC-IC assay, each proliferation experiment used three cell subsets (MNC, CD133+, and Lin− cells) separated from the same UCB unit in order to rule out interpatient variability. Over 8 weeks in liquid culture, both CD133+ and Lin− cells proliferated significantly better and showed higher CFC output under TPOFLK synergism when compared with K36EG, which was not optimized for Lin− cell expansion. Our data further confirmed that K36EG cytokine mix recruits HSPC into proliferation and differentiation, whereas TPOFLK synergism induced HSPC self-renewal and maintenance, making the latter combination more adequate for ex vivo expansion studies [8, 52, 53]. However, our data also suggest that to maintain CFC-erythroid output longer when expanding Lin− cells, TPOFLK synergism might be optimized by adding EPO after 2 weeks of liquid culture.
Lin− cells proliferated significantly less rapidly and less extensively than CD133+ cells upon TPOFLK synergism when considering the total cumulative CFC production. However, expanded Lin− cells showed a significantly higher CFC frequency than CD133+ cells at weeks 4, 6, and 8 in TPOFLK-stimulated liquid culture. This could be explained by HSPC asymmetric divisions and/or by primitive LTC-IC having a slower proliferation potential than the more mature CFC and more abundant precursor cells in the CD133+ cell subset [19, 38]. Therefore, as the Lin− cell population in this study contained more LTC-IC but fewer CFC than CD133+ cells, these cells underwent slower proliferation upon TPOFLK synergism, while possibly expanding a more primitive HSPC pool than the CD133+ cell subset. This may be of particular importance, as a recent study by Kuci et al.  demonstrated that a discrete CD34−CD133+ HSPC subset, despite producing no CFC in a clonogenic assay, exhibited the highest severe combined immunodeficient repopulating cell frequency reported in the literature so far. Therefore, although CD133+ cells demonstrated higher proliferation potential under TPOFLK synergism, expanded Lin− cells may be a more appropriate cell source for repopulation assays. This is currently under investigation in our laboratory.
This study describes a standardized, cost-reduced negative isolation protocol for purification of very primitive Lin− cells. In UCB, this Lin− cell population was highly enriched in LTC-IC and demonstrated slow HSPC expansion properties in TPOFLK-stimulated liquid culture.