Ex Vivo Expansion Does Not Alter the Capacity of Umbilical Cord Blood CD34+ Cells to Generate Functional T Lymphocytes and Dendritic Cells


  • Ladan Kobari,

    1. Laboratoire d'Hématologie, Unité de Formation et de Recherche EA1638, Université Pierre et Marie Curie, CHU Saint Antoine, Paris, France
    Search for more papers by this author
  • Marie C. Giarratana,

    1. Laboratoire d'Hématologie, Unité de Formation et de Recherche EA1638, Université Pierre et Marie Curie, CHU Saint Antoine, Paris, France
    Search for more papers by this author
  • Jean C. Gluckman,

    1. UMR 7151 CNRS - Université Paris 7 and Laboratoire d'Immunologie Cellulaire et Immunopathologie de l'Ecole Pratique des Hautes Etudes, Institut Universitaire d'Hématologie, Hôpital Saint-Louis, Paris, France
    Search for more papers by this author
  • Luc Douay M.D.,

    Corresponding author
    1. Laboratoire d'Hématologie, Unité de Formation et de Recherche EA1638, Université Pierre et Marie Curie, CHU Saint Antoine, Paris, France
    2. Service d'Hématologie Biologique, Hôpital Armand Trousseau, Assistance Publique Hôpitaux de Paris, France
    • Service d'Hématologie Biologique, Hôpital Armand Trousseau, 26 avenue du Docteur Arnold Netter, 75012 Paris, France. Telephone: +33-1-44-73-62-22; Fax: +33-1-44-73-63-33
    Search for more papers by this author
  • Michelle Rosenzwajg

    1. Laboratoire d'Hématologie, Unité de Formation et de Recherche EA1638, Université Pierre et Marie Curie, CHU Saint Antoine, Paris, France
    2. UMR 7087 CNRS - Université Pierre et Marie Curie, Hôpital de la Pitié-Salpêtrière, Paris, France
    Search for more papers by this author


We examined whether ex vivo expansion of umbilical cord blood progenitor cells affected their capacity to generate immune cells such as T lymphocytes (TLs) and dendritic cells (DCs). The capacity to generate TLs from cord blood CD34+ cells expanded for 14 days (d14) was compared with that of nonexpanded CD34+ cells (d0) using fetal thymus organ cultures or transfer into nonobese diabetic/severe combined immunodeficient mice. The cell preparations yielded comparable percentages of immature (CD4+CD8, CD4+CD8+) TLs and functional mature (CD3+CD4+, CD3+CD8+) TLs with an analogous TCR (T-cell receptor)-Vβ repertoire pattern. As regards DCs, d0 and d14 CD34+ cells also yielded similar percentages of CD1a+ DCs with the same expression levels of HLA-DR, costimulatory and adhesion molecules, and chemokine receptors. DCs derived from either d14 or d0 CD34+ stimulated allogeneic TLs to the same extent, and the cytokine pattern production of these allogeneic TLs was similar with no shift toward a predominant Th1 or Th2 response. Even though the intrinsic capacity of d14 CD34+ cells to generate DCs was 13-fold lower than that of d0 CD34+ cells, this reduction was offset by the prior amplification of the CD34+ cells, resulting in the overall production of 15-fold more DCs. These data indicate that ex vivo expansion of CD34+ cells does not impair T lymphopoiesis nor DC differentiation capacity.


There is a wealth of literature on the expansion of hematopoietic stem/progenitor cells (HPCs) as a means to improve the efficacy of hematopoietic cell transplantation. If amplification of the myeloid compartment has been abundantly dealt with, few data regarding lymphopoeisis are available, and the immune competence of amplified cells has not yet been examined. Therefore, we addressed the question of whether ex vivo expansion of umbilical cord blood (UCB) CD34+ cells changes the characteristics of the immune cells that are subsequently generated relative to those obtained from nonmanipulated HPCs.

UCB HPCs have been demonstrated to possess significant advantages over their bone marrow or peripheral blood counterparts in terms of proliferative capacity and immunological reactivity [1]. The latter observation may be accounted for by the naive nature of circulating neonatal T lymphocytes (TLs) and/or the presence of peripheral tolerance-inducing TLs or dendritic cells (DCs) [2, [3], [4]–5]. However, the low number of HPCs in the UCB is a key limitation for transplantation, which could be overcome by ex vivo expansion using growth factors acting on early HPCs [6, [7], [8], [9], [10], [11]–12]. In this context, we have reported clinically relevant culture conditions that support the self-renewal of immature cells retaining long-term myeloid and lymphoid potentials in the nonobese diabetic/severe combined immunodeficient (NOD/SCID) mouse model, together with ongoing telomerase activity that preserves telomere ends [8]. The expanded products do not contain mature T, B, or natural killer (NK) cells but conserve progenitors capable of long-term lymphoid reconstitution [8].

Although it is well known that UCB cells have the capacity to generate all types of mature lymphoid and myeloid cells [8, 13, [14], [15]–16] and that expanded cell preparations thereof display efficient myelopoiesis, the capacity of the latter to differentiate into mature competent immune cells has not been fully investigated.

Therefore, the aim of the present study was to determine the incidence, if any, of the expansion methodology on the immune competence of an expanded potential graft. Because we have already shown the capacity of expanded CD34+ cells to differentiate into mature and functional NK cells [17], we assessed here the quantitative and qualitative characteristics of the immune cells (i.e. TLs and DCs) that differentiated after expansion. Because antigen-specific TL responses depend on T-cell receptor (TCR) recognition of antigenic peptides presented by major histocompatibility complex molecules on antigen-presenting cells (in particular DCs), we examined whether ex vivo expansion of UCB CD34+ cells allowed mature and functional TLs to subsequently generate without skewing their TCR-Vβ chain expression patterns. Because DCs, as the most potent antigen-presenting cells, are key regulators of the immune system, DCs generated from expanded CD34+ cells were also examined for their differentiation, phenotype, and function.

Materials and Methods

Isolation of UCB CD34+ Cells

UCB units from normal full-term deliveries were obtained from Saint-Vincent de Paul Hospital (Paris) with the mothers' informed consent. Mononuclear cells (MNCs) were separated by Ficoll-Isopaque centrifugation (Seromed, Biochrom, Berlin, http://www.seromed.com). CD34+ cells were isolated by supermagnetic microbead selection using a high-gradient magnetic field and two runs on Mini-MACS columns (Miltenyi Biotech GmbH, Bergisch Gladbach, Germany, http://www.miltenyibiotec.com), which yielded 90% ± 2.5% CD34+ cells (n = 9). After 14-day ex vivo expansion, the cell preparations were purified in the same manner, resulting in 92% ± 3% CD34+ cells (n = 9). These CD34+ cell preparations are referred to hereafter as “d0” and “d14”, respectively.

CD34+ Cell Expansion

CD34+ cells (104 cells/ml) were cultured in serum-free long-term culture medium (LTCM) [18] with the following recombinant human growth factors [8]: 100 ng/ml Flt3-ligand (FL), 100 ng/ml megakaryocyte growth and development factor (MGDF) (both from Abcys, Paris, http://www.abcys.fr), 100 ng/ml stem cell factor (SCF), and 10 ng/ml granulocyte colony-stimulating factor (G-CSF) (both from R&D Systems Europe Ltd, Abingdon, Oxfordshire, U.K., http://www.rndsystems.com). According to our established protocol [19], the cells incubated at 37°C in humidified 5% CO2 were seeded in 4 ml of complete LTCM. On day 6, 16 ml of fresh medium with cytokines was added to the flasks. After 14 days, cells were washed in Iscove-modified Dulbecco's medium (Seromed), counted by trypan blue exclusion, analyzed for HPCs, and immunophenotyped. Cell expansion was expressed as fold increases (i.e. the absolute number of d14 cells divided by that of d0 cells).

HPC and Long-Term Culture-Initiating Cell Assays

HPC and long-term culture-initiating cell (LTC-IC) assays were performed as described [8]. Colony-forming units-granulocytes/macrophages (CFU-GM) and burst-forming units-erythrocytes (BFU-E) were assayed in 0.9% methylcellulose, and colonies with more than 50 cells were scored on culture day 14. LTC-ICs were maintained for 5 weeks on murine stromal MS5 cells (generous gift of Kazuhiro Mori, Niigata University, Ikarashi, Japan). LTC-IC frequencies were determined by scaling cultures down to 100 μl in 96-well microtiter plates (Falcon; BD Biosciences Discovery Labware, Bedford, MA, http://www.bdbiosciences.com) and performing limiting-dilution assays with 20–50 replicates per dilution step (2–64 cells/well on day 0 and 32–1 × 104 cells/well on day 14). Absolute LTC-IC counts corresponded to the cell concentrations, yielding 37% negative wells using Poisson statistics.


NOD-LtSz-scid/scid (NOD-SCID) breeding pairs were originally obtained from John Dick (University Health Network, Toronto, Canada). Mice were raised under sterile conditions in containers equipped with air filters in the animal facilities of Institut Gustave Roussy (Villejuif, France). Females pregnant for 14–16 days were used in the fetal thymic organ culture (FTOC) experiments, the age of embryos being calculated from day 0, the morning on which the mating plug was observed.

Human TL Development Assay in FTOCs

FTOCs were performed as reported [16]. Briefly, each NOD-SCID fetal thymic lobe was incubated for 2 days in RPMI 1,640, 10% human AB serum (Institut Jacques Boy, Tours, France), 5% fetal calf serum (FCS), 100 IU/ml penicillin, 2 mM l-glutamine (all from Invitrogen, Carlsbad, CA, http://www.invitrogen.com), plus 5 ng/ml interleukin (IL)-2, 20 ng/ml IL-7 (both from Diaclone, Besançon, France, http://www.diaclone.com/anglais), and 50 ng/ml SCF (complete medium). An aliquot of 2–17 × 104 d0 or d14 CD34+ cells was then added to each well. After 48 hours, the thymic lobes were transferred to filters floating in complete medium without cytokines in 6-well plates (Falcon). After 6 weeks, cells were recovered from the lobes by mechanical dissociation and then pooled, and human cells were stained for flow cytometric analysis.

NOD/SCID Mouse Repopulation Assay and Assays of TL Function

Prior to transplantation, 6-week-old NOD/SCID mice were given a sublethal 3.5-Gy whole body irradiation (cobalt-60 Eldorado S irradiator; AECL Medical, Mississauga, Ontario, Canada, http://www.aecl.ca/index.asp) and were injected intraperitoneally with 200 μg of TM-β1, an antibody functionally blocking the mouse IL-2Rβ chain (kindly provided by Jean Plum, University Hospital Ghent, Ghent, Belgium) [20]. After 24 hours, the mice were anesthetized briefly during injection into the retro-orbital vein with either d0 (n = 25) or d14 (n = 25) human CD34+ cells (1 × 105 to 5 × 105 cells/mouse). Fifteen weeks after injection, the mice were killed, and thymus or spleen cells were recovered. In each condition (n = 25), thymus or spleen cells were pooled, and human CD3+ TL were isolated using MACS (magnetic microbead cell-sorting) and used for phenotypic analysis and functional assays.

The proliferation assay was conducted on sorted CD3+ TL as reported [20]. Briefly, 5 × 104 CD3+ cells per well harvested were incubated for 3 weeks in the presence of 2 μg/ml phytohemagglutinin (PHA) (Sigma-Aldrich, St. Louis, http://www.sigmaaldrich.com), 50 U/ml recombinant human IL-2 and irradiated human peripheral blood mononuclear cells. The expanded cells were then incubated in 96-well, flat-bottomed, microtiter plates (Falcon) and stimulated or not (controls) for 24 hours with 10 μg/ml anti-CD3 monoclonal antibody (mAb) (OKT3; American Type Culture Collection, Manassas, VA, www.atcc.org) plus 10 ng/ml 12-O-tetradecanoylphorbol 13-acetate (TPA) (Sigma-Aldrich). For CD3 stimulation, plates were coated for 1 hour at 37°C with 100 μl/well 10 μg/ml purified OKT3 mAb in phosphate-buffered saline (PBS). After 24 hours of stimulation, cultures were centrifuged; supernatants were harvested and stored at −20°C before being assayed for interferon (IFN)-γ production by enzyme-linked immunosorbent assay (BD Biosciences PharMingen, San Diego, http://www.bdbiosciences.com/pharmingen/products).

DC Cultures

DCs were differentiated from CD34+ cells as described [14]: d0 or d14 CD34+ cells (5 × 104/ml) were cultured in six-well plates (Falcon; BD Biosciences Discovery Labware), at 37°C in humidified 5% CO2, in RPMI 1640, 10% FCS, 1% l-glutamine, 1% penicillin/streptomycin (Invitrogen), 10 mM sodium pyruvate, 50 ng/ml SCF, 10 ng/ml tumor necrosis factor (TNF)-α (the latter two from R&D Systems, Minneapolis), 50 ng/ml FL, and 20 ng/ml granulocyte-macrophage colony-stimulating factor (GM-CSF) (R&D Systems). Cultures were split on day 5 and then every 3–4 days, and the cells were phenotypically analyzed on day 12.

Flow Cytometry Analysis

Cells were washed in PBS, 2% FCS, incubated for 30 minutes at 4°C with the appropriate mAbs, washed again, and fixed in PBS, 1% paraformaldehyde, before analysis on an EPICS XL4C flow cytometer (Beckman Coulter, Miami, http://www.beckmancoulter.com). Acquisitions were performed using the System II software, and analysis with the EXPO32 ADC software. Cells were permeabilized with the Intraprep solution (Beckman Coulter) for intracellular labeling.

TLs were labeled with fluorescein isothiocyanate (FITC)-, phycoerythrin (PE)-, phycoerythrin-cyanin-, and/or PE-Texas Red (ECD)-conjugated mAbs against CD3, CD4, CD8, and CD45 (Beckman Coulter, Villepinte, France), using appropriate isotype-matched control mAbs. TCR-Vβ expression was investigated with a panel of 24 TCR-Vβ-specific mAbs (Beta Mark TCR-Vβ repertoire kit, Beckman Coulter) used in association with CD3 and CD45 labels according to the manufacturer's instructions. The TL repertoire of UCB MNCs was examined in parallel as a control.

DCs were studied with FITC-, PE-, and/or ECD-conjugated mAbs against the following antigens: CD11b, CD11c, CD18, CD34, CD40, CD45, CD58, CD80, CD83, CD86 (Beckman Coulter), CD1a, CD14, CD54, CXCR4, CCR5, mannose receptor (MR), and HLA-DR (BD Biosciences, San Jose, CA, http://www.bdbiosciences.com). Controls were isotype-matched irrelevant mAbs from BD Biosciences or Beckman Coulter.

Allogeneic Mixed Leukocyte Reaction Assay

The assay was performed in culture medium, 10% heat-inactivated normal human AB serum. Allogeneic TLs (1 × 105/well) were cultured for 6 days in 96-well culture microplates (Falcon) as responders to 1–1 × 104 DCs generated from d0 or d14 CD34+ cells [14]. [3H]Thymidine incorporation (specific activity 1 Ci/mM; GE Healthcare, Little Chalfont, Buckinghamshire, U.K., http://www.gehealthcare.com) was measured after a 1 μCi per well 8- to 16-hour pulse. The results are shown as mean cpm of triplicates; incorporation in negative control wells, with responder TLs alone, was never greater than 1,000 cpm.

For assaying cytokine production, stimulating DCs (1 × 105) generated from d0 or d14 CD34+ cells were cocultured in 24-well plates (Corning Life Sciences, Acton, MA, http://www.corning.com/lifesciences) with allogeneic responder TLs (1 × 106) in 1.5 ml of medium, 10% heat-inactivated human AB serum. Supernatants were collected after 48 hours and assessed for IL-2, IL-4, IL-5, IL-10, and IFN-γ production using a Cytometric Bead Array (CBA) kit (Becton, Dickinson and Company).


Statistical analyses were performed with the Mann-Whitney U test for the TCR-Vβ repertoire and Student's t test for all other experiments. p values less than .05 were considered statistically significant.


Ex Vivo Expansion of CD34+ UCB Cells

On the basis of our previous studies of UCB HPC ex vivo expansion [8, 19, 21], UCB CD34+ cells were expanded by 14-day culture under stroma-free conditions in serum-free medium containing SCF, FL, MGDF, and G-CSF. Under these conditions, expansion was 1,497 ± 67-fold for total cells, 91 ± 15 for CD34+ cells, 121 ± 19 for CFU-GM (known to comprise DC progenitors [22]), 14 ± 3 for BFU-E, and 8 ± 1 for LTC-IC. These results confirmed our findings that d14 CD34+ cells had hematopoietic properties comparable with those of nonexpanded CD34+ cells and that the final product was not modified by the expansion procedure [8, 19, 21].

NOD-SCID FTOCs Support T-Cell Differentiation of Expanded CD34+ Cells

Six weeks after seeding FTOCs with d0 or d14 CD34+ cells (n = 5) [23, 24], the human lymphoid cells recovered were identified on the basis of strong CD45 expression and low side scatter (SSClow). Among these (Fig. 1), most expressed intracytoplasmic CD3 (iCD3), confirming their T-lymphoid origin, and percentages of CD4+ cells generated from d0 and d14 cells were 82% ± 4% and 73% ± 10%, respectively. Cells at different stages of maturation were also found in comparable proportions, with 31% ± 8% double-positive (DP) iCD3+CD4+CD8+ cells in both cases whereas simple-positive (SP) iCD3CD4+CD8 cells, which comprise a majority of immature TLs prior to the DP stage, represented 50% ± 7% and 42% ± 6% of CD45+ cells, respectively. In line with another report [25], less than 10% of cells in both preparations were mature SP CD4+CD8 or CD4CD8+ TLs (data not shown).

Figure Figure 1..

Lymphoid progenitors are present in umbilical cord blood d0 and d14 CD34+ cells. Fetal thymic organ cultures were seeded with d0 CD34+ cells or d14 CD34+. After 6 weeks, cells were analyzed by flow cytometry. T lymphocytes were identified among CD45+ low side scatter cells according to their CD4, CD8, and iCD3 expression. Data are from one experiment of five. Abbreviations: d0, nonexpanded; d14, expanded for 14 days; iCD3, intracytoplasmic CD3.

Phenotypic and Functional Analysis of Human TLs Derived from Expanded CD34+ Cells

Because the FTOC model did not permit us to obtain sufficient numbers of mature CD4+ and CD8+ TLs [26], another strategy was used to confirm the capacity of expanded CD34+ cells to differentiate into mature, functional TLs.

NOD/SCID mice pretreated with an anti-IL2 mAb were injected with d0 or d14 CD34+ cells (n = 25) as described [20, 27]. After 15 weeks, thymus and spleen cells were cultured for 3 weeks with PHA plus IL-2 before analysis. Human cells derived from d0 CD34+ cells were gated in CD45+SSClow cells, 64% of which were mature TLs expressing membrane (m)CD3+, 28% being CD3+CD4+, 4% CD3+CD8+, and 32% CD3+CD4 CD8; similar data were obtained with d14 CD34+ cells, 78% being mCD3+ cells, 46% CD3+CD4+, 17% CD3+CD8+, and 15% being CD3+CD4 CD8 (Fig. 2). These data confirm the capacity of d14 CD34+ cells to differentiate into mature TLs to the same extent as fresh d0 CD34+ cells.

Figure Figure 2..

Phenotype of human mature TLs recovered from the nonobese diabetic/severe combined immunodeficient mouse repopulation assay. d0 and d14 CD34+ cells were injected into TM-β1-treated mice. After 15 weeks, sorted human CD3+ cells were seeded with interleukin-2, phytohemagglutinin, and irradiated human peripheral blood mononuclear cells. Human TLs were analyzed after 3 weeks of stimulation. Abbreviations: d0, nonexpanded; d14, expanded for 14 days; TL, T lymphocyte.

To assess and compare the function of the TLs derived from d0 and d14 CD34+ cells, these cells were first stimulated with PHA, followed by OKT3 mAb and TPA, and their capacity to produce IFN-γ upon stimulation was examined. Table 1 shows that TLs derived from both d0 and d14 CD34+ cells could produce equally high amounts of IFN-γ, confirming the capacity of the expanded cell preparations to give rise to functional TLs.

Table Table 1.. IFN-γ capacity of human T lymphocytes derived from expanded d0 and d14 CD34+ cells
original image

A mAb panel that distinguishes 24 TCR-Vβ families, covering ∼70% of the human repertoire, was then used to compare these mCD3+ TLs generated from d0 and d14 CD34+ cells to determine whether CD34+ cell ex vivo expansion permitted development of a normal TL repertoire. Results were compared with those obtained from five different UCB MNC samples as controls. Although TCR-Vβ expression patterns varied among the different cell preparations, they did not significantly differ as regards the repertoires of TLs from the UCB and those derived from either d0 or d14 CD34+ cells, and there was no evidence of abnormal skewing in favor of particular Vβ chains, all being represented in each case (Fig. 3).

Figure Figure 3..

TCR-Vβ repertoire of mCD3+ TLs generated from d0 and d14 CD34+ cells. (A): Comparison of the TCR-Vβ repertoires of UCB MNCs and d0 CD34+ cell-derived TLs. (B): Comparison of the TCR-Vβ repertoires of d0 and d14 CD34+ cell-derived TLs. Isotype control antibodies were included in each staining series, and results are expressed as percentages of cells above background. Data represent means ± SEM from five experiments. Abbreviations: d0, nonexpanded; d14, expanded for 14 days; TCR, T-cell receptor; TL, T lymphocyte; UCB MNC, umbilical cord blood mononuclear cell.

Differentiation of DCs from Expanded CD34+ Cells

To examine their DC potential, we cultured d0 and d14 CD34+ cells with SCF, FL, GM-CSF, and TNF-α, as reported [13, [14]–15]. Because the DCs in these cultures develop as CD1a+CD14 and CD1a+CD14+ populations [13, [14]–15], analysis was primarily based on their CD1a and CD14 expression, which showed no significant difference between d0 and d14 CD34+ cell-derived DCs beyond culture day 5, with similar percentages of CD1a+ (whether CD14+ or CD14) DCs (35% ± 9% vs. 37% ± 6%, n = 5) on day 12. Furthermore, whatever the initial cell preparation, the DCs had the same phenotype and expressed the same level of MR, HLA-DR, costimulatory molecules (CD40, CD80, CD86), adhesion molecules (CD11b, CD11c, CD18, CD54, CD58), chemokine receptors (CXCR4, CCR5), and mature DC marker CD83 cells (Fig. 4) [14, 15, 28]. Thus, d0 and d14 CD34+ cells differentiated to the same extent into DCs having a comparable phenotype.

Figure Figure 4..

Phenotype of CD1a+ dendritic cells generated from d0 and d14 CD34+ cells. Cells were gated on CD1a expression. Gray histograms: negative control; solid histograms: staining with the relevant monoclonal antibody. Data are from one experiment of five. Abbreviations: d0, nonexpanded; d14, expanded for 14 days.

Quantitatively, when the numbers of DCs derived from d0 or d14 CD34+ cells were related to CD34+ input cell numbers from either day 0 or day 14, yields appeared to be lower from expanded than from fresh progenitors: the intrinsic capacity of d14 cells to generate DCs was then 13 ± 4-fold less than that of d0 cells (p = .02, n = 5). However, this reduction was offset by the prior amplification of d14 CD34+ cells, the numbers of which were on the average 91-fold higher than those of d0 CD34+ cells, which finally led to the generation of 15 ± 5 more DCs (p = .03, n = 5) (Fig. 5).

Figure Figure 5..

Effect of ex vivo expansion on DC production. (A): Numbers of DCs recovered per d0 or d14 CD34+ cell. (B): Total numbers of DCs recovered per initial d0 CD34+ cell subsequently expanded or not. Data are means ± SEM from five experiments. Abbreviations: d0, nonexpanded; d14, expanded for 14 days; DC, dendritic cell.

Effect of Ex Vivo Expansion on the Capacity of DCs to Stimulate and Polarize TL Responses

Finally, we compared the ability of DCs derived from d0 and d14 CD34+ cells to stimulate TLs by coculturing graded amounts of DCs with allogeneic TLs for 6 days. Under these conditions, DCs generated from d14 CD34+ cells stimulated the TLs to the same extent as DCs derived from fresh CD34+ cells (Fig. 6). In parallel, cytokine production by these stimulated allogeneic TLs was assessed in 48-hour supernatants. As expected, cytokine release from DCs or TLs cultured alone was very low, whereas TLs cocultured with DCs of either origin produced comparable cytokine patterns without any shift toward a predominant Th1 or Th2 response, as indicated by the combined production of IL-2, IFN-γ, as well as IL-4 and IL-10 (Fig. 6).

Figure Figure 6..

DC capacity to stimulate and polarize allogeneic TLs. Allogeneic TLs cultured as responder cells with DCs originating from d0 or d14 CD34+ cells. (A): The proliferative response was assessed by [3H]thymidine incorporation (cpm). (B): The cytokine production was measured in supernatants. Data are from one experiment of three. Abbreviations: cpm, counts per minute; d0, nonexpanded; d14, expanded for 14 days; DC, dendritic cell; IFN, interferon; IL, interleukin; TL, T lymphocyte; TNF, tumor necrosis factor.

Discussion and Conclusion

Much interest is currently focused on improving in vitro methods to expand HPCs without compromising their original development and engrafting potential. We have reported an ex vivo expansion protocol that supports the self-renewal of transplantable HPCs and the amplification of myeloid progenitors while maintaining the lymphoid potential of the graft [8]. However, almost no differentiated lymphoid cells or DCs were identified among the cells harvested after 14 days of expansion. In the present work, we first confirmed that subsequent culture of such expanded UCB CD34+ cells under TL or DC differentiation conditions led to the production of TL precursors, mature TLs [16, 24], and DCs. We then examined whether ex vivo expansion of CD34+ cells modified the immune competence of their progenies. This is a particularly important point because alterations of TL functions or repertoire could induce increased alloreactivity of the expanded graft or abnormal immune responses. On the other hand, because DCs are the most potent antigen-presenting cells, it was also important to verify that the expanded cells could differentiate into DCs able to initiate immune responses that were not skewed toward an undesirable polarization. Here, we investigated only myeloid DC differentiation because, in humans, in vitro plamascytoid DC differentiation protocols from cord blood are not yet readily achievable from cord blood HPCs [29].

T lymphopoiesis was first assessed using the FTOC assay, which retains the architecture of the thymic environment and can support development of human TLs from their progenitors. In this environment, human progenitors differentiate into CD4+CD8+ DP intermediate cells but only very few reach the CD4+CD8 or CD4CD8+ SP mature TL stage [8, 26]. To obtain mature TLs and to investigate their function, we injected CD34+ cells into NOD/SCID mice treated with anti-IL2, which is known to limit NK development [20]. TL development in FTOCs seeded with d14 or d0 CD34+ cells was similar with respect to the proportions of TL precursors [24, 26, 30, 31], and mature CD4+CD8 and CD4CD8+ SP TLs were obtained from both CD34+ cell preparations after transfer into NOD/SCID mice. Upon TCR stimulation by OKT3 and IL-2, these mature TLs produced high IFN-γ amounts, indicating that TLs differentiated from expanded CD34+ cells are as functional as TLs derived from nonmanipulated CD34+ cells. We next determined whether the TCR repertoire of TLs differentiated from d14 CD34+ cells was conserved relative to TLs derived from d0 CD34+ cells and to control UCB MNCs. Flow cytometry analysis of the distribution of Vβ families [32, 33] showed that TLs originating from d14 or d0 CD34+ cells indeed expressed all TCR-Vβ families with no gross skewing toward any particular Vβ group, suggesting that the TL repertoire was conserved [32].

Therefore, ex vivo expansion of UCB CD34+ cells does not appear to notably alter the differentiation, function, or TCR-Vβ diversity of the TLs they generate, and it should permit the development of normal TLs with the potential to recognize a wide range of immunogens.

In parallel, we investigated the capacity of expanded CD34+ cells to differentiate into DCs. Progress in the understanding of DC development from progenitor cells has resulted from the use of in vitro culture systems to generate large numbers of DCs from CD34+ HPCs [13, [14]–15]. After 12 days of culture in the presence of SCF, FL, GM-CSF, and TNF-α, d0 and d14 CD34+ cells yielded similar percentages of CD1a+ DCs, expressing functionally important molecules (HLA-DR, costimulatory molecules, adhesion molecules, and chemokines receptors) to the same extent. Thus, d0 and d14 CD34+ cells differentiated in the same manner into DCs with a comparable phenotype.

DCs are the most potent antigen-presenting cells and can polarize TL responses. Th1 TLs play an important role in graft-versus-host disease (GVHD) by secreting IL-2 and IFN-γ, which can induce cytotoxic TL responses [34]. Conversely, IL-4 and IL-10 produced by Th2 cells prevent IFN-γ secretion and may suppress GVHD. Polarization of TL responses toward Th1 or Th2 depends on the type of antigen-presenting cells that contact the TLs during their activation, and in this context, DCs have been shown to play an important role in GVHD [35]. Here, DCs differentiated from d14 CD34+ cells displayed the same functional characteristics as DCs derived from nonmanipulated CD34+ cells. When allogeneic TLs were stimulated with DCs derived from either CD34+ preparation, the same levels of proliferation and the same pattern of cytokine production were observed, suggesting that ex vivo expansion does not affect the capacity of the resulting DCs to stimulate or polarize TLs.

UCB cell transplantation is associated with a low incidence of GVHD, possibly due to TL immaturity and also to a higher proportion of peripheral tolerance-inducing TLs or DCs, especially plasmocytoid DCs (pDCs) [3, 5]. Whereas the pDC/myeloid DC (mDC) ratio in the UCB has been shown to be greater than in postnatal blood [3], no cells corresponding to blood mDCs or pDCs were detected after expansion in the CD34 cell fraction (data not shown), but our methodology was not designed to induce or amplify pDCs and in vitro-generated mDCs are phenotypically different from blood mDCs.

Whereas increased frequency of GVHD was reported in recipients of expanded UCB cells using another expansion protocol [36], with our expansion method we did not find changes in TL function and repertoire or in DC phenotype or capacity to stimulate or polarize TLs as compared with TLs or DCs derived from d0 CD34+ cells, which might account for this increased frequency [36]. However, it was not possible to investigate the other side of the coin (i.e. increased alloreactivity).

The only difference we found concerned absolute numbers of DCs obtained after amplification. The intrinsic capacity of d14 CD34+ cells to generate DCs was decreased, but because of the CD34+ cell amplification before differentiation into DC, we finally obtained an overall production of approximately 15-fold more DCs than without amplification. Arrighi et al. used another CD34+ cell expansion system with SCF, FL, and thrombopoietin (TPO) and subsequent culture with GM-CSF and TNF-α to induce DC differentiation [37], and they found that there was no further cell growth during DC culture after expansion. In our system, we obtained both CD34+ cell amplification and cell growth, which resulted in a final yield of approximately 700 DCs per fresh initial CD34+ cell. No correlation between the number of CD34+ cells transplanted in vivo and DCs reconstitution after HPC transplantation has been shown, but larger quantities of CD34+ progenitors in the graft have been associated with improved survival and decreased transplant-related mortality [38]. In the context of UCB transplantation, there is a putative interest to infuse expanded cells. This could permit the increase of numbers of HPCs and committed precursors in the graft, which might result in more rapid engraftment and accelerated hematopoietic reconstitution without subverting immune reconstitution [3, 39].


The authors indicate no potential conflicts of interest.


We thank J. Plum and B. Vandekerckhove for the gift of the TM-β1 antibody and the interesting discussion, P. Ardouin and A. Rouchès for their invaluable help in the breeding and care of the NOD-SCID mice, and Van Nifterik (Saint-Vincent de Paul Hospital) for kindly providing the cord blood samples. We also thank A. M. Megiovanni and C. Menard for their technical assistance. This work was supported by grants from the Fondation de France and the Association Combattre la Leucémie.