Long-Term Ex Vivo Maintenance and Expansion of Transplantable Human Hematopoietic Stem Cells

DEVELOPMENT of ex vivo culture conditions that facilitate in vitro maintenance and expansion of long-term transplantable hematopoietic stem cells (HSCs) is a crucial component and a major challenge in stem cell research. This is a necessary first step toward a better understanding of the regulatory process governing the development of all hematopoietic lineages from HSCs. Furthermore, establishment of such culture conditions is a prerequisite for potential ex vivo manipulation and expansion of transplantable HSCs in several clinical applications such as gene therapy, tumor cell purging, and stem cell transplantation.

Several clinical transplantation studies have shown that human hematopoietic cells can be cultured ex vivo with or without stroma and still retain the capacity to engraft human recipients, although it is not known whether the number of transplantable HSCs has changed.1-3 Furthermore, all of these studies involve autologous transplantation, making it difficult to determine whether long-term repopulation was derived from cultured cells or from surviving endogenous stem cells. Whether primitive human hematopoietic cells are actually expanded during ex vivo culture is controversial because different assays and culture conditions have been used in different studies.4-6 It has recently been shown that highly purified subfractions of CD34⁺ cells possess the greatest proliferative potential, resulting in a large expansion of colony-forming cells (CFC), while long-term culture-initiating cells (LTC-IC) show either a slight reduction7 or moderate increase.6 Although these in vitro assays provide an essential quantitative assessment of the functional properties of the expanded cells, they do not evaluate repopulation capacity. Appropriate assays are needed and must be used for the direct quantitation of the most primitive and transplantable stem cells.

HSCs are defined as cells having both capacity for self-renewal and ability to differentiate into at least 8 distinct hematopoietic cell lineages. Hematopoietic progenitors in human bone marrow (BM) can be identified by the expression of the CD34 antigen. However, the majority of the CD34⁺ BM cells are committed to specific hematopoietic cell lineages. Using different assays8 for the properties expected from HSC, enrichment for pluripotent progenitor cells in the CD34⁺ cell fraction has been successfully performed by eliminating the CD34⁺ cells expressing lineage-associated antigens,9,10 HLA-DR,10CD38,11 CD45RA,12 or CD71,10,12 or lacking thy-1.13,14 Because none of these assays can provide an accurate determination of the HSC identity, and no single assay has been used across the board to compare different HSC phenotypes, the utility of these assays is still being debated.

In the mouse, no in vitro system has been developed that adequately recapitulates stem cell behavior, and, currently, the only reliable functional assay system for the most primitive stem cell compartment is long-term in vivo transplantation. Experimental in vivo hematopoietic assays are not ethical with human subjects. Studies of human stem cell renewal, differentiation, and maintenance would be facilitated by the availability of a relevant animal model in which assays similar to those used for secondary transfer and long-term reconstitution of mice can be employed. Over the past decade, several groups have transplanted human hematopoietic cells into immunodeficient mouse strains in an attempt to develop a relevant and reproducible in vivo transplantation model. A number of in vivo assays for human HSC are now available.15-18 However, in these systems, the only hematolymphoid microenvironments in which the human cells can differentiate are those of murine origin. Because of the complexity of the regulatory mechanisms surrounding hematopoiesis, a murine-origin microenvironment may not be physiologically sufficient to sustain human stem cell self-renewal and differentiation. This might explain why large doses of exogenous human cytokines,15,17 high numbers of unfractionated input cells,15,17 and cotransplantation with stromal cells engineered to produce human cytokines16are needed to sustain human hematopoiesis in these murine systems. Recent data suggest that the nonobese diabetic severe combined immunodeficiency (NOD/SCID) mouse is a better host than the SCID mouse for this type of research.18-20 Physiologically, it would be preferable to sustain human HSCs in a hematopoietic microenvironment of human origin in a murine model. To address this problem, a “humanized” murine model, the SCID-hu mouse, was developed by implantation of intact human hematolymphoid tissues into SCID mice.21 Studies have shown that the SCID-hu mouse is a useful and relevant in vivo system for assaying the developmental potential of transplantable human HSCs and that assays similar to those used for secondary transfer and long-term reconstitution of mice can be employed.13,22 It is notable that lymphoid development from primitive cells is better demonstrated in the SCID-hu mouse model.13,22 

Recently, a SCID repopulating cell (SRC) assay19 has been used to perform a quantitative assessment of the repopulation capacity of ex vivo–cultured cells initiated with CD34⁺CD38⁻ cells.23 This study showed a 4- and 10-fold increase in the number of CD34⁺CD38⁻ cells and CFC, respectively, as well as a 2- to 4-fold increase in SRC after 4 days of culture.23 However, after 9 days of culture, all SRC were lost, despite increases in total cells, CFC counts, and CD34⁺ cells.23 These results support the notion that appropriate quantitative assays for transplantable stem cells are essential for the development of culture conditions that support primitive cells.23 In this study, our strategy is the following: First, we would like to establish a quantitative assay based on in vivo reconstitution in the SCID-hu mouse model to quantify the number of transplantable human stem cells. This assay will be used to measure the number of stem cells before and after culture from the same donors. Second, we would like to develop a culture system based on a murine BM-derived stromal cell line and exogenous cytokines. Here we report the development of a novel culture system for facilitating ex vivo maintenance and expansion of transplantable human CD34⁺ thy-1⁺ cells. In this ex vivo culture system, the absolute number of cells with CD34⁺ thy-1⁺ phenotype increases 150-fold after 5 weeks of culture. Those ex vivo–expanded CD34⁺thy-1⁺ cells not only maintain their CD34⁺thy-1⁺ phenotype, but also preserve their engrafting ability for long-term hematopoietic reconstitution in the SCID-hu mice. Furthermore, we have identified that leukemia inhibitory factor (LIF) is the responsible cytokine for maintaining and expanding those transplantable CD34⁺ thy-1⁺ cells in this ex vivo culture system and that LIF exerts its role in expanding transplantable CD34⁺ thy-1⁺ cells by indirectly affecting the stromal AC6.21 cells. Our results also suggest that ex vivo expansion of transplantable human stem cells under this in vitro culture system is a general phenomenon and not just specific for CD34⁺ thy-1⁺ cells.

Human fetal bone, thymus, and liver tissues were dissected from 18- to 24-week-old fetuses obtained by elective abortion with approved consent (Anatomic Gift Foundation, White Oak, GA). A sample of each received fetal tissue was stained with a panel of monoclonal antibodies (MoAbs) to HLA to establish the donor allotype. The fetal tissues were used either for construction of the SCID-hu mice or for preparation of human HSCs. To purify human HSCs, BM cell suspensions were prepared by flushing split long bones with RPMI 1640 (GIBCO/BRL, Gaithersberg, MD) containing 2% heat-inactivated fetal calf serum (FCS; Gemini Bio-Products, Inc, Calabasas, CA). Low-density (<1.077 g/mL) mononuclear cells were isolated (Lymphoprep; Nycomed Pharma, Oslo, Norway) and washed twice in staining buffer (SB) consisting of Hanks’ Balanced Salt Solution (HBSS) with 2% heat-inactivated FCS and 10 mmol/L HEPES. Samples were then incubated for 10 minutes with 1 mg/mL heat-inactivated human gammaglobulin (Gamimune; Miles Inc, Elkhart, IN) to block Fc receptor binding of mouse antibodies. Fluorescein isothiocyanate (FITC)-labeled CD34 MoAbs and phycoerythrin (PE)-labeled thy-1 MoAbs (or PE-labeled CD38 MoAbs) were then added at 0.5 to 1 μg/10⁶ cells in 0.1 to 0.3 mL SB for 20 minutes on ice. Control samples were incubated in a cocktail of FITC-labeled and PE-labeled isotype-matched MoAbs. Cells were washed twice in SB, and then resuspended in SB containing 1 μg/mL propidium iodide (Molecular Probes Inc, Eugene, OR) and sorted using the tri-laser fluorescence-activated cell sorter MoFlo (Cytomation, Inc, Fort Collins, CO). Live cells (ie, those excluding propidium iodide) were always greater than 95%. Sort gates were set based on the mean fluorescence intensity of the isotype control sample. Cells were collected in 12- or 24-well plates in RPMI 1640 containing 10% FCS and 10 mmol/L HEPES, counted, and reanalyzed for purity in every experiment. Typically, 450,000 to 500,000 CD34⁺thy-1⁺ cells were obtained from a single donor. MoAbs for CD34 and CD38 were purchased from Becton Dickinson (Mountain View, CA). MoAbs for thy-1 and isotype controls were purchased from Pharmingen (San Diego, CA).

Sorted cells were cultured on a preestablished monolayer of mouse stromal cell line AC6.21. Briefly, 5 × 10³ to 1 × 10⁴ stromal cells were plated in 96-well flat-bottom plates 1 week before the experiment in 100 μL of long-term culture medium (LTCM) consisting of RPMI 1640, 5 × 10⁻⁵ mol/L 2-mercaptoethanol, 10 mmol/L HEPES, penicillin (50 U/mL) and streptomycin (50 mg/mL), 2 mmol/L sodium pyruvate, 2 mmol/L glutamine, and 10% FCS. Twenty CD34⁺thy-1⁺ cells were distributed in 100 μL of LTCM into each well in 96-well flat-bottom plates with preestablished AC6.21 monolayer. Individual growth factor or combinations of growth factors were added immediately after seeding the sorted cells into the microtiter plates at a concentration of 10 ng/mL of each growth factor. Human recombinant interleukin-3 (IL-3), IL-6, granulocyte-macrophage colony-stimulating factor (GM-CSF), stem cell factor (SCF), and LIF were purchased from R&D Systems (Minneapolis, MN). Half of the culture medium was replaced weekly with fresh LTCM containing the respective growth factor(s). To minimize disturbance to the cultures during the weekly medium change, 100 μL of old medium from each well was removed slowly from the top of the well with a multiple channel pipetter, and 100 μL of fresh medium was then slowly added to each well.

The extent of each growth factor or combinations of growth factors to support in vitro expansion of purified human fetal BM stem cells was determined weekly by counting the total number of hematopoietic cells present in 15 individual wells in each culture. For counting the content of hematopoietic cells in these wells, cells were harvested individually from these wells by gently pipetting off the wells without destruction of the stromal layer. At the end of the culture period, either week 5 or week 7, individual wells were analyzed for lineage content by flow cytometry. Cells from each well were harvested individually by vigorously pipetting of the wells including stromal cells. Half of the cells from each well were stained with FITC- or PE-labeled MoAbs against CD19 and CD33, and the other half were stained with antibodies against CD34 and thy-1 or CD38. Analysis was gated on the hematopoietic cells, excluding the stromal cells, and the quadrants were set based on the mean fluorescence intensity of the isotype control samples. FITC- or PE-labeled MoAbs against CD19 and CD33 were purchased from Pharmingen (San Diego, CA). Cells were analyzed on a FACScan fluorescent cell analyzer (Cytomation, Inc, Fort Collins, CO). To purify the ex vivo–expanded HSCs from the culture, cells from each well were harvested individually by vigorously pipetting off the wells including stromal cells at the end of the culture. One-twentieth of the cells (200,000 cells per well, 1/20 fraction of each well contains about 10,000 cells) from each well was divided into 2 fractions. One fraction was stained with antibodies against CD34 and thy-1 or CD38, and another fraction was stained with FITC- or PE-labeled MoAbs against CD19 and CD33 and analyzed on the FACScan cell analyzer. The cells from the wells containing all 3 populations, including CD33⁺ cells, CD19⁺ cells, and CD34⁺thy-1⁺ cells or CD34⁺ CD38⁻cells, were pooled together (usually about 10% of the wells in the cultures initiated with CD34⁺thy-1⁺ cells and 12% of the wells in the cultures initiated with CD34⁺ CD38⁻ cells) and sorted for HSCs, either CD34⁺ thy-1⁺ or CD34⁺ CD38⁻ cells, as described above.

C.B-17 scid/scid mice were bled in our facility under sterile conditions. Mice used for human tissue transplantation were 6 to 8 weeks of age, and the construction of SCID-hu thymus/liver (thy/liv) and bone model mice were constructed as previously described13,22 and in accordance with the guidelines set forth by the City of Hope Research Animal Care Committee. At the time of surgery, animals were weighed and anesthetized with a mixture of ketamine (50 mg/kg) and xylazine HCL (25 mg/kg) administered intraperitoneally. For thy/liv mice, individual pieces (1 to 2 mm) of human fetal thymus and autologous liver were placed under the kidney capsule of C.B-17 scid/scid mice and allowed to engraft for 3 months before stem cell reconstitution. For bone model mice, pieces of fetal bone were placed subcutaneously and allowed to vascularize for 2 to 3 months. Animals were preconditioned by total body irradiation with 350 rads 4 to 6 hours before they were subjected to stem cell reconstitution. The ability of the purified human fetal BM HSCs, including CD34⁺ thy-1⁺ and CD34⁺CD38⁻ populations (stem cell donor is always selected to be HLA-MA2.1–positive), either fresh uncultured or ex vivo–expanded, to reconstitute thymus and BM was tested by direct inoculation into irradiated grafts (thy/liv and bone, the graft is always selected to be HLA-MA2.1–negative). A limiting dilution experiment was conducted to determine quantitatively the transplantable cells in the freshly purified human fetal BM CD34⁺thy-1⁺ population. Four different cell doses were evaluated (10,000, 3,000, 1,000, and 300 cells in 10 μL of HBSS were injected per human graft). For other reconstitution experiments, 10,000 cells were used because 10,000 CD34⁺ thy-1⁺ cells purified from fresh fetal BM reproducibly establish long-term hematopoietic reconstitution in greater than 90% of SCID-hu mice in our laboratory. Control animals were injected with HBSS only. Engraftment was analyzed at 3 to 4 months postinjection. Human bones were removed and split open to flush the marrow cavity with SB. Collected cells were spun down and the pellet was resuspended for 5 minutes in a red blood cell lysing solution. Cells were washed twice in SB and counted before being stained for 2-color immunofluorescence with directly labeled MoAbs against HLA allotypes in combination with CD19 and CD33. Human thymus grafts were recovered, reduced to a cellular suspension, and subjected to 2-color immunofluorescence analysis using directly labeled MoAbs against HLA allotypes in combination with CD3, CD4, and CD8. FITC- and PE-conjugated irrelevant mouse Igs were used as negative controls. Cells were analyzed on a FACScan fluorescent cell analyzer. FITC- or PE-labeled CD19, CD33, CD3, CD4, and CD8 were purchased from Pharmingen (San Diego, CA).

Limiting dilution experiments were performed to determine if the in vivo reconstitution assay in SCID-hu mice is capable of quantifying transplantable stem cells in the heterogeneous CD34⁺thy-1⁺ population. CD34⁺ thy-1⁺cells (HLA-MA2.1–positive) were sorted from 22- to 24-week-old fetal BM samples and 4 different cell doses (10,000, 3,000, 1,000, and 300 cells per graft) were evaluated. Thirty SCID-hu mice (15 thy/liv and 15 bone mice) were used for each cell dose, and a total of 120 mice were used for the CD34⁺ thy-1⁺ cells purified from each fetal BM sample. The ability of the injected cells to engraft the SCID-hu mice was determined by flow cytometry 3 months after cell injection. Four different fetal BM samples were used and compared in these experiments (Table 1). For donor no. 1, donor reconstitution derived from freshly purified CD34⁺thy-1⁺ cells was evident in 87% (13 of 15), 20% (3 of 15), 7% (1 of 15), and 0% (0 of 15) of the bone grafts and in 93% (14 of 15), 20% (3 of 15), 7% (1 of 15), and 0% (0 of 15) of the thymic grafts from an injected cell dose of 10,000, 3,000, 1,000, and 300, respectively. The percentage of donor-derived cells in the bone grafts of reconstituted animals was 41% ± 10% (ranging from 30% to 54%), 9% ± 3% (ranging from 6% to 12%), 2.2% from an injected cell dose of 10,000, 3,000, and 1,000, respectively. The percentage of donor-derived cells in the thymic grafts of reconstituted animals was 50% ± 8% (ranging from 40% to 58%), 12% ± 4% (ranging from 8% to 16%), and 3.2% from a injected cell dose of 10,000, 3,000, and 1,000, respectively. The data from the other 3 donors were almost identical to donor no. 1 and showed no significant donor effect in this assay.

To determine the appropriate cell dose that can provide a quantitative measurement for transplantable stem cells for subsequent in vivo reconstitution assay in SCID-hu mice, calibration curves for cell doses and reconstitution rates in both thy/liv and bone models were established by logistic regression based on the data from limiting dilution experiments (Table 1). The calibration curves for both thy/liv and bone models are almost identical (Fig1). These curves suggest that cell doses from 1,000 to 10,000 could be used to quantify the number of transplantable stem cells with a 50% reconstitution rate at approximately 5,000 cells. Because the 10,000-cell dose gives rise to a higher reconstitution rate of about 90% and the percentage of donor-derived cells in the reconstituted animals only reaches 40% to 50%, this suggests that the 10,000-cell dose is not at the saturation level yet in this in vivo reconstitution assay. We have decided to use 10,000 cells as the standard cell dose for subsequent in vivo transplantation experiments. This assay was then used to quantify the number of transplantable CD34⁺ thy-1⁺ cells, before and after culture, from the same donors.

The present study was designed to develop a simple culture system to facilitate in vitro maintenance and expansion of transplantable stem cells. Our strategy was to use cloned stromal cell lines because the stromal cell line offers several advantages over the heterogeneous adherent cell monolayer derived from BM.24 Several stromal cell lines have been shown to support in vitro myelopoiesis,25-28 B lymphopoiesis,28 or in some cases both.29-31 We have performed single-cell (CD34⁺ thy-1⁺) sorting experiments in the AC6.21 stromal cell line, derived from the Whitlock-Witte culture system,32 coculture assay. Individual CD34⁺thy-1⁺ cells were deposited onto a preestablished AC6.21 monolayer without the addition of exogenous cytokines. Our unpublished data suggest that 1 in 20 CD34⁺ thy-1⁺ cells was able to initiate long-term cultures and become growth-positive (>100 progeny), and 50% of these growth-positive wells gave rise to both B lymphocytes and myeloid cells. Our unpublished results are similar to a previously published report using a different stromal cell line, SyS-1.13 Because the AC6.21 stromal cell line can provide an environment for a single multipotential CD34⁺thy-1⁺ cell to differentiate into both B-lymphocyte and myeloid cells, a phenomenon similar to the in vivo BM environment, we hypothesized that it might also provide a natural environment for a primitive stem cell to self-renew, which in turn will facilitate the development of an ex vivo culture system for HSC expansion. We have decided to use AC6.21 as a stromal support in developing an ex vivo culture system. Based on our own results from single-cell coculture experiments, we have decided to initiate the cultures with 20 purified human fetal BM CD34⁺ thy-1⁺ cells in each well and anticipated that 100% of the wells would be growth-positive in 5 weeks. From the single-cell coculture experiments, our results showed that coculture with AC6.21 was not able to maintain cells with a CD34⁺ thy-1⁺ phenotype in the culture. To evaluate whether the addition of cytokines to the culture will help to maintain the cells with stem cell phenotype in this coculture system, IL-3,33,34 IL-6,33 GM-CSF,35SCF,34,36,37 and LIF38,39 were selected as exogenous cytokines based on their potential for expanding hematopoietic progenitor cells from in vitro studies.

We evaluated the effects of the 5 individual cytokines on the proliferative capacity of CD34⁺ thy-1⁺ cells in AC6.21 stroma cocultures. Kinetic studies (Fig 2) showed no obvious proliferation for the first 2 weeks in culture, regardless of the treatments. This was followed by a continuous increase in total cell numbers, which reached about 5,000 cells per well by week 7. By week 7, almost all wells (>95%) for each treatment scored as growth-positive (>5,000 cells/well) as expected for CD34⁺ thy-1⁺ cells from our own unpublished data and a previously published report.13 However, the culture treated with LIF showed a late-phase exponential growth from the fourth week and reached about 100,000 cells per well by week 7. This result suggests that among the 5 cytokines evaluated, LIF is the only cytokine by itself that can facilitate the proliferation of purified CD34⁺thy-1⁺ cells. Even though IL-3, IL-6, GM-CSF, and SCF, individually, cannot promote cell growth in this coculture system compared with the control (without exogenous cytokine), they were able to establish synergistic effects on HSC expansion with LIF. Kinetic studies (Fig 3) show the synergistic effects of LIF with other cytokines on the proliferative capacity of purified human CD34⁺ thy-1⁺ cells. The addition of other cytokines to LIF accelerates the proliferative kinetics of purified human fetal BM CD34⁺thy-1⁺cells, and these cultures enter an exponential phase of growth from the second week in culture (v the fourth week with LIF alone). At the end of the fifth week in culture, more than 95% of the wells for all treatments except LIF alone are confluent (about 200,000 cells per well). However, it is noteworthy that those cultures treated with combinations of LIF and other cytokines still remain quiescent during the first week in culture (v 2 weeks with LIF alone).

To assess the effects of these cytokine combinations on the differentiation potential of purified CD34⁺thy-1⁺ cells, cells from individual wells were analyzed by flow cytometry for the presence of CD33⁺ myeloid cells and CD19⁺ B lymphocytes. Approximately 50% of the wells in each treatment were mixed lymphoid/myeloid wells, ranging from 41% of the GM-CSF–treated culture alone to 62% of those treated with LIF + IL-3 + IL-6 + GM-CSF + SCF (Table 2). The percentage of CD33⁺ myeloid cells within these mixed lymphoid/myeloid wells ranges from 40% of those treated with LIF + IL-3 + IL-6 to 55% of those treated with LIF + IL-3 + IL-6 + GM-CSF; and the percentage of CD19⁺ cells ranges from 15% of those treated with combinations of LIF with other cytokines to 8% of the control group (Table 2). The frequency of mixed lymphoid/myeloid wells and the percentages of the myeloid and B-cell populations within these mixed lymphoid/myeloid wells among these different treatments are very similar to the control. These results suggest that addition of cytokine combinations in this coculture system does not dramatically alter the differentiation potential of purified human fetal BM CD34⁺thy-1⁺ cells.

To determine if addition of exogenous cytokines to the coculture system is capable of facilitating the maintenance and expansion of cells with CD34⁺ thy-1⁺ phenotype, flow cytometric analyses were performed to detect and quantify the number of CD34⁺ thy-1⁺ cells in each individual well in these cultures. As expected from single-cell coculture experiments, cells with CD34⁺ thy-1⁺ phenotype were not detectable in the control group (Table 3). Among the 5 cytokines evaluated, LIF is the only factor by itself that can give rise to CD34⁺ thy-1⁺/positive wells. The percentage of CD34⁺ thy-1⁺ cells in these positive wells is about 7%. Because each well was initiated with 20 cells and only about 10% of the wells are CD34⁺thy-1⁺/positive, this suggests that the frequency of cells capable of regenerating CD34⁺ thy-1⁺ phenotype is about 1 in 200 within the CD34⁺ thy-1⁺population. The frequency of CD34⁺thy-1⁺/positive wells in those cultures treated with combinations of LIF with other cytokines ranges from 10% of those treated with LIF + IL-3 + IL-6 + GM-CSF to 12% of those treated with LIF + IL-3 + IL-6 + GM-CSF + SCF (Table 3). All CD34⁺thy-1⁺/positive wells, regardless of treatments, contain both CD33⁺ and CD19⁺ cells. The percentage of CD34⁺ thy-1⁺ cells within the CD34⁺thy-1⁺/positive wells increases significantly from 7% in LIF-treated cultures to 15% in cultures treated with combinations of LIF and other cytokines (Table 3; P < .00001). As there are about 200,000 cells per well within these CD34⁺thy-1⁺/positive wells in these cultures, the absolute number of CD34⁺ thy-1⁺ cells in each well averages about 30,000, representing a 1,500-fold expansion in this population during the 5 weeks in culture (each well was initiated with 20 purified CD34⁺ thy-1⁺ cells). The overall bulk equivalent (ie, not counting only the 10% to 12% of wells containing CD34⁺ thy-1⁺ cells) is in excess of a 150-fold expansion of CD34⁺ thy-1⁺ cells under these culture conditions. Although the difference in the frequency of CD34⁺ thy-1⁺/positive wells in these cultures is not statistically significant (P = .85), cultures treated with LIF + IL-3 + IL-6 + GM-CSF + SCF had a higher frequency (12%) of CD34⁺thy-1⁺/positive wells (Table 3) and were selected as the optimal condition for subsequent experiments.

To determine whether ex vivo–expanded CD34⁺thy-1⁺ cells possess the same in vivo proliferative and differentiation capacity as freshly purified human fetal BM CD34⁺ thy-1⁺ cells, the ability of these cells to engraft and initiate long-term hematopoietic reconstitution in SCID-hu mice was analyzed by multiparameter flow cytometry. A representative analysis of hematopoietic reconstitution in the SCID-hu mouse transplanted with 10,000 ex vivo–expanded CD34⁺thy-1⁺ cells is shown in Fig 4. The thymic engrafting potential of ex vivo–expanded CD34⁺thy-1⁺ cells is shown in Fig 4A. The engrafted human thymus of this SCID-hu mouse contained 50% ex vivo–expanded CD34⁺ thy-1⁺–derived thymocytes as detected by expression of the donor HLA (MA2.1-positive). These cells were further analyzed for their expression of T-cell markers CD3, 4, and 8, and they displayed a normal T-cell maturation pattern. The BM engrafting potential of ex vivo–expanded CD34⁺ thy-1⁺cells is shown in Fig 4B and C. The engrafted human bone fragment of this SCID-hu mouse contained 39% donor-derived CD19⁺ B cells (Fig 4B) and 16% donor-derived CD33⁺ myeloid cells (Fig 4C) as detected by donor HLA (MA2.1-positive). The control mice did not receive any donor cells and did not show any detectable donor-derived cells. These results show that ex vivo–cultured and expanded CD34⁺ thy-1⁺ cells not only maintain their CD34⁺ thy-1⁺ phenotype, but also preserve their in vivo engrafting potential.

To further compare the number of transplantable CD34⁺thy-1⁺ cells before and after culture, we have compared the in vivo engrafting activity between freshly purified and ex vivo–expanded CD34⁺ thy-1⁺ cells from the same donors. Data from three independent experiments using cells from different donors were compiled (Table 4). As expected from previous limiting dilution experiments with 10,000 freshly purified CD34⁺ thy-1⁺ cells (Table 1), all 3 donors gave rise to about a 90% reconstitution rate in both thy/liv and bone mice with 40% to 50% donor-derived cells in the reconstituted animals (Table 4). Ex vivo–expanded CD34⁺thy-1⁺ cells from the same donors gave almost identical results as the freshly purified CD34⁺ thy-1⁺cells in both the frequency of reconstitution and the percentage of donor-derived cells in the reconstituted animals (Table 4). These results suggest that CD34⁺ thy-1⁺ cells, before and after culture, are similar, both qualitatively and quantitatively. In this set of experiments, we have also evaluated whether the ex vivo–expanded hematopoietic cells from wells without detectable CD34⁺ thy-1⁺ cells (about 90% of the wells in this culture are CD34⁺ thy-1⁺/negative wells) are capable of engrafting in SCID-hu mice. Our results show that donor reconstitution was not detectable (0 of 60) in the SCID-hu mice when transplanted with 10,000 ex vivo–expanded hematopoietic cells from CD34⁺ thy-1⁺/negative wells for all 3 donors, suggesting that transplantable cells, before and after culture, are only present in the CD34⁺ thy-1⁺ population.

To directly quantify the number of transplantable CD34⁺thy-1⁺ cells before and after culture, the in vivo engraftment data (Table 4) were subjected to statistical analysis. The calibration curves established from limiting dilution experiments (Fig1) were used to estimate an equivalent dose of fresh cells based upon the engraftment data derived from CD34⁺ thy-1⁺cells before and after culture from the same donors. Separate logistic regression curves were estimated for thy/liv and bone models, as shown in Fig 5A and B, respectively. In both cases, the estimation of equivalent dose of fresh CD34⁺thy-1⁺ cells for 10,000 cultured CD34⁺thy-1⁺ cells was larger than 10,000 cells (13,600 for the thy/liv model, and 16,350 for the bone model). The lower (95%) confidence bounds on equivalent fresh CD34⁺thy-1⁺ cells for 10,000 cultured cells were 7,295 for the thy/liv model and 7,905 for the bone model.40 These results suggest that the ex vivo–expanded CD34⁺ thy-1⁺cells have higher in vivo engrafting potential than the freshly purified CD34⁺ thy-1⁺ cells from the same donors, and imply that there is a preferential expansion of transplantable CD34⁺ thy-1⁺ cells rather than just the cells with the CD34⁺ thy-1⁺ phenotype under this culture system.

Our data have shown that LIF plays key roles in facilitating maintenance and expansion of transplantable human CD34⁺thy-1⁺ cells under this culture system. To understand how LIF mediates its effects on maintenance and expansion of transplantable human CD34⁺ thy-1⁺ cells under this ex vivo culture system, our experimental approach was to determine whether LIF acts directly on human CD34⁺ thy-1⁺ cells or indirectly on the murine AC6.21 stromal cells. It has been previously reported that human LIF (huLIF) can bind to the murine receptor and is biologically active in mouse M1 cells.41 However, there is no binding of murine LIF (muLIF) to the human receptor.42This published information provided us with the rationale for using muLIF in our culture system. If huLIF acts directly on human CD34⁺ thy-1⁺ cells to promote their ex vivo expansion, then muLIF will not be able to serve the same function because muLIF cannot bind to human CD34⁺ thy-1⁺cells.41,42 However, if huLIF exerts its roles indirectly by acting on AC6.21 stromal cells, then muLIF will be able to substitute for huLIF in facilitating ex vivo expansion of human CD34⁺ thy-1⁺ cells.41,42 Results from muLIF experiments demonstrate that muLIF performs its function as efficiently as huLIF in facilitating maintenance and expansion of CD34⁺ thy-1⁺ cells in vitro (Table 5). These results suggest that LIF facilitates maintenance and expansion of CD34⁺thy-1⁺ cells in this ex vivo culture system not by directly acting on human CD34⁺ thy-1⁺ cells but by indirectly affecting AC6.21 stromal cells. This conclusion is further supported by the previous results from kinetic studies (Figs 2 and 3). The lack of proliferative response of CD34⁺thy-1⁺ cells for the first 1 to 2 weeks in culture suggests that these cells are waiting for the AC6.21 stroma to establish a suitable microenvironment, which in turn is induced by the presence of LIF.

To determine if the activity to facilitate ex vivo expansion of transplantable stem cells by this in vitro culture system is a general phenomenon and not only specific to CD34⁺thy-1⁺ cells, we decided to investigate the proliferative responses of another HSC candidate (CD34⁺CD38⁻ population)11 under this ex vivo culture system. Since we have established a quantitative assay for transplantable stem cells based on the SCID-hu mouse model, our first set of experiments was to demonstrate that CD34⁺CD38⁻ cells are capable of establishing long-term hematopoietic reconstitution in SCID-hu mice. A representative analysis of hematopoietic reconstitution in SCID-hu mice transplanted with 10,000 freshly purified CD34⁺ CD38⁻ cells is shown in Fig 6. The thymic engrafting potential of CD34⁺ CD38⁻ cells is shown in Fig 6A. The engrafted human thymus of this SCID-hu mouse contained 30% donor-derived thymocytes as detected by expression of the donor HLA (MA2.1-positive). These cells also displayed a normal T-cell maturation pattern. The BM engrafting potential of fresh CD34⁺ CD38⁻ cells is shown in Fig 6B and C. The engrafted human bone fragment of this SCID-hu mouse contained 18% donor-derived CD19⁺ B cells and 15% donor-derived CD33⁺ myeloid cells as detected by donor HLA (MA2.1-positive). In contrast, donor-derived cells were not detected in the control mice transplanted with 10,000 CD34⁺CD38⁺ cells. These results suggest that there is an enrichment for the in vivo engrafting activity in the CD34⁺CD38⁻ population compared with the CD34⁺CD38⁺ population. Our results showed that about 90% of the SCID-hu mice, both bone and thy/liv mice, are reconstituted by injecting 10,000 freshly purified human fetal BM CD34⁺CD38⁻ cells into each human graft (data not shown).

To facilitate a relevant comparison between CD34⁺CD38⁻ population and CD34⁺thy-1⁺ population, the same experimental procedures were followed. The kinetic studies of cultures initiated with CD34⁺ CD38⁻ cells show an early wave of proliferation in the first 2 weeks that was absent in the cultures initiated with the CD34⁺ thy-1⁺ population (Fig 7). This was followed by a second wave of exponential growth from the end of the second week to the fourth week, and by the fourth week most of the wells were confluent (Fig 7). These results show that the growth response of CD34⁺CD38⁻ cells under this ex vivo culture system is 1 week faster than and distinguishable from the CD34⁺thy-1⁺ population.

To determine the differentiation potential of purified CD34⁺ CD38⁻ cells under this ex vivo culture system, cells from individual wells were analyzed by flow cytometry for the presence of CD33⁺ myeloid cells and CD19⁺ B lymphocytes. The phenotypic analyses of 5-week cultures derived from 3 different donors are summarized in Table 6. The frequency of mixed lymphoid/myeloid wells in 5-week cultures is 12% (36 of 300), 10% (31 of 300), and 11% (34 of 300) for donors no. 21, 22, and 23, respectively. This frequency of mixed lymphoid/myeloid wells derived from CD34⁺ CD38⁻ cells (Table 6) is significantly lower (P < .00001) than the respective 50% from the CD34⁺ thy-1⁺ population (Table 2). We have estimated that about 1 in 182 CD34⁺CD38⁻ cells is multipotential for both B-cell and myeloid lineages (v 1 in 40 CD34⁺thy-1⁺ cells) under this ex vivo culture condition. It is also notable that the percentage of CD19⁺ B lymphocytes (Table 6) is significantly lower (P < .00001) in these cultures, averaging 8%, compared to 15% for the cultures initiated with CD34⁺ thy-1⁺ cells (Table 2). However, the percentage of CD33⁺ myeloid cells (Table 6) is significantly higher (P < .0001) in these cultures, averaging 64%, compared to 45% for the cultures initiated with CD34⁺ thy-1⁺ cells (Table 2).

To determine the maintenance and expansion potential of purified CD34⁺ CD38⁻ cells under this ex vivo culture system, flow cytometric analyses were performed on the 5-week cultures to quantify the number of CD34⁺CD38⁻ cells in the wells of these cultures. The frequency of CD34⁺CD38⁻-positive wells in these cultures is 69% (207 of 300), 71% (214 of 300), and 72% (216 of 300) for donors no. 21, 22, and 23, respectively (Table 7). This result suggests that 1 in 29 CD34⁺ CD38⁻ cells (v 1 in 200 CD34⁺ thy-1⁺ cells) is capable of expansion under this ex vivo culture condition. The percentage of CD34⁺ CD38⁻ cells in myeloid wells (CD34⁺CD38⁻/positive wells without detectable CD19⁺ cells), averaging 9%, is significantly lower (P < .00001) than the respective percentage, averaging 15%, in the mixed lymphoid/myeloid wells. All of the mixed lymphoid/myeloid wells (36 wells from donor no. 21, 31 wells from donor no. 22, and 34 wells from donor no. 23) contain CD34⁺CD38⁻ cells. These data show that 1 in 29 CD34⁺ CD38⁻ cells is capable of proliferation, but only 1 in 182 CD34⁺CD38⁻ cells is capable of both proliferation and multipotential differentiation. The higher frequency and percentage of myeloid wells in these cultures are evidence suggesting that the majority of CD34⁺ CD38⁻ cells which are capable of proliferation in the cultures have already committed to the myeloid lineage. However, the CD34⁺ CD38⁻cells in those mixed lymphoid/myeloid wells present both expected characteristics of HSC, proliferation, and multipotential differentiation. In those mixed lymphoid/myeloid wells, an average of 15% (30,000 cells) are CD34⁺ CD38⁻cells, representing a 1,500-fold expansion of cells with the CD34⁺ CD38⁻ phenotype. The overall bulk equivalent (ie, not counting only the 10% to 12% of mixed lymphoid/myeloid wells) is in excess of a 150-fold expansion of CD34⁺ CD38⁻ cells under this culture system.

Our next experiment was to determine whether ex vivo–expanded CD34⁺ CD38^– cells still maintain their ability to establish long-term hematopoietic reconstitution in SCID-hu mice, and to determine whether ex vivo–expanded CD34⁺CD38⁻ cells from myeloid wells and mixed lymphoid/myeloid wells differ in their in vivo engrafting activity. The ex vivo–expanded CD34⁺ CD38⁻ cells were sorted from pools of myeloid wells and mixed lymphoid/myeloid wells, respectively, derived from 5-week cultures. The sorted cells in HBSS were then injected (10,000 cells/graft) into human thymus and bone fragments in SCID-hu mice. Control mice were injected with HBSS only. Thymic and BM engraftment by CD34⁺ CD38⁻cells in the SCID-hu mice were analyzed 3 to 4 months after stem cell injection. Data from 3 independent experiments using cells from different donors were compiled (Table 8). Donor reconstitution derived from 10,000 freshly purified CD34⁺ CD38⁻ cells was evident in about 90% of the bone grafts and 90% of the thymic grafts for all 3 donors. These results are consistent with our unpublished data and are also similar to the reconstitution rate when 10,000 CD34⁺thy-1⁺ cells were injected. The percentage of donor-derived hematopoietic cells was about 35% in both the bone and thymic grafts for all 3 donors (Table 8). Compared with the percentage of donor-derived cells in the animals transplanted with 10,000 CD34⁺ thy-1⁺ cells (40% in the bone mice and 50% in the thy/liv mice), the percentage of donor-derived cells from CD34⁺ CD38⁻ cells is lower. However, ex vivo–expanded CD34⁺ CD38⁻ cells purified from mixed lymphoid/myeloid wells give the same results in both reconstitution rate and percentage of donor-derived cells as freshly purified CD34⁺ CD38⁻ cells from the same donors (Table 8). These results suggest that ex vivo–expanded CD34⁺ CD38⁻ cells from mixed lymphoid/myeloid wells are similar, both qualitatively and quantitatively, to freshly purified CD34⁺CD38⁻ cells from the same donors. In contrast, reconstitution was only detected in 10% (3 of 30) of the bone grafts and 0% (0 of 30) of the thymic grafts with ex vivo–expanded CD34⁺ CD38⁻ from myeloid wells. These in vivo data, which are consistent with in vitro results obtained from CD34⁺ CD38⁻ cultures, suggest that although CD34⁺ CD38⁻ cells in the myeloid wells are capable of regenerating CD34⁺CD38⁻ cells, the majority of these cells have already committed to the myeloid lineage and are not able to differentiate into lymphoid lineage, including T and B lymphocytes. Because ex vivo–expanded CD34⁺ CD38⁻ cells give almost the same reconstitution rate and percentage of donor-derived cells as freshly purified CD34⁺ CD38⁻cells from the same donors, it is reasonable to conclude that transplantable CD34⁺ CD38⁻ cells have been similarly expanded 1,500-fold as the cells with CD34⁺CD38⁻ phenotype in these mixed lymphoid/myeloid wells. The bulk equivalent (ie, the overall culture, not just the 10% to 12% of mixed lymphoid/myeloid wells) is in excess of a 150-fold expansion.

Efforts to develop culture systems for the maintenance of transplantable stem cells have met with limited success for the last 2 decades. One major problem in developing an ex vivo culture system for HSC expansion has been finding an assay to quantitatively measure transplantable stem cells before and after culture. It has been repeatedly shown that combinations of cytokines can exert potent stimulatory effects on stem/progenitor populations. Although in vitro studies are often used to study stem cell biology, it is difficult to relate these results to what occurs in the in vivo setting. Many studies have shown that progenitor cells with myeloid-erythroid and lymphoid potential can be expanded in vitro; however, the in vivo transplantation potential of these ex vivo–expanded progenitor cells remains obscure. In this study, we have first established a quantitative assay for transplantable human stem cells using in vivo reconstitution in SCID-hu mice as the model system. Limiting dilution experiments in SCID-hu mice using 4 different fetal BM samples (Table1) have established calibration curves for both thy/liv and bone models by logistic regression (Fig 1). There was no significant donor effect and no significant lack of fit in either case, and the calibration curves are suitable baselines for quantitative measurement of transplantable cells in a heterogeneous stem cell population. These calibration curves also suggest that cell doses ranging from 1,000 to 10,000 used in establishing the calibration curves can be employed in the assay to relatively quantify the number of transplantable CD34⁺ thy-1⁺ cells in the heterogeneous CD34⁺ thy-1⁺ population. This quantitative assay system was used to quantify the number of transplantable human stem cells before and after culture.

In this study, we used CD34⁺ thy-1⁺ cells from human fetal BM as input HSC phenotype. The CD34⁺thy-1⁺ population is very heterogeneous. Studies from many groups have found that the CD34⁺ thy-1⁺phenotype does not exclusively select primitive stem cells only. The CD34⁺ thy-1⁺ population contains a proportion of cells able to repopulate SCID-hu mice,13,22 a proportion of extended LTC-IC (ELTC-IC),4 a large number of LTC-IC,6 a large number of CFC,7 and even some more differentiated cells.13 In this study, the 1,500-fold expansion in the CD34⁺ thy-1⁺/positive wells is based on the phenotype of output cells subsequent to culture. Because the CD34⁺ thy-1⁺ population is heterogeneous, it remains possible that the 1,500-fold expansion might represent the expansion of those more differentiated cells. To address whether the proportion of cells capable of engrafting SCID-hu mice is similarly expanded 1,500-fold as CD34⁺ thy-1⁺ cells after culture, the engraftment potential of ex vivo–expanded CD34⁺ thy-1⁺ cells was determined in SCID-hu mice. If the 1,500-fold expansion of CD34⁺thy-1⁺ cells in the CD34⁺thy-1⁺/positive wells is largely contributed by the other cells and not by the proportion of cells capable of engrafting SCID-hu mice, then we would expect to see a decrease in the frequency of reconstituted SCID-hu mice injected with 10,000 ex vivo–expanded CD34⁺ thy-1⁺ cells compared to the mice injected with 10,000 freshly purified CD34⁺thy-1⁺ cells. Our data (Table 4) from a comparative analysis of 296 SCID-hu mice transplanted with CD34⁺thy-1⁺ cells before and after culture show that the proportion of CD34⁺ thy-1⁺ cells capable of engrafting SCID-hu mice and the percentage of donor-derived hematopoietic cells in the reconstituted animals remain the same before and after culture (Table 4). Because the whole CD34⁺thy-1⁺ population in the CD34⁺thy-1⁺/positive wells has expanded 1,500-fold over the 5-week culture period and the proportion of cells capable of engrafting SCID-hu mice remains the same within the CD34⁺thy-1⁺ phenotype before and after culture, these results suggest that the proportion of cells capable of engrafting SCID-hu mice in the CD34⁺ thy-1⁺/positive wells has been similarly expanded 1,500-fold. Statistical analysis based on the calibration curves established by logistic regression from limiting dilution experiments (Table 1) further supports that the equivalent dose of fresh CD34⁺ thy-1⁺ cells for 10,000 cultured CD34⁺ thy-1⁺ cells was larger than 10,000 cells (13,600 for the thy/liv model, Fig 5A; and 16,350 for the bone model, Fig 5B). These results suggest that there is a preferential expansion of transplantable CD34⁺ thy-1⁺ cells under this ex vivo culture system. Because we are not able to identify a single cell type responsible for long-term engraftment in SCID-hu mice, the experimental results presented here are not yet able to unequivocally prove that the proportion of cells capable of engrafting SCID-hu mice in the CD34⁺ thy-1⁺/positive wells is expanded 1,500-fold as the CD34⁺ thy-1⁺phenotype over the 5-week culture period.

To rule out the possibility that this in vitro culture system might be specific for CD34⁺ thy-1⁺ cells and can only facilitate the expansion of CD34⁺ thy-1⁺ cells, we have investigated the proliferative responses of another human HSC candidate, CD34⁺ CD38⁻ cells, under this in vitro culture condition. Because CD34⁺CD38⁻ cells have a different phenotype than CD34⁺ thy-1⁺ cells, these cells show similar but distinguishable growth kinetics compared with CD34⁺thy-1⁺ cells. The kinetic studies demonstrate that cultures initiated with the CD34⁺ CD38⁻ cells show an early wave of proliferation in the first 2 weeks followed by a second wave of exponential growth from the end of second week to the fourth week, and that the overall cell growth in the cultures initiated with CD34⁺ CD38⁻ cells is 1 week faster than the cultures derived from CD34⁺ thy-1⁺population (Fig 7). Under this ex vivo culture system, the frequency of cells in the CD34⁺ CD38⁻ population capable of multipotential differentiation for B-cell and myeloid lineages is estimated to be 1 in 182, which is significantly lower (P < .00001) than the 1 in 40 for CD34⁺thy-1⁺ cells (Tables 2 and 6). The frequency of cells in the CD34⁺ CD38⁻ population capable of proliferation is estimated to be 1 in 29, which is significantly higher (P < .00001) than the 1 in 200 for CD34⁺thy-1⁺ cells (Tables 3 and 7). These data suggest that the majority of CD34⁺ CD38⁻ cells that are capable of proliferation in the cultures have already committed to the myeloid lineage, which might account for the higher frequency and percentage of myeloid wells in these cultures. However, the frequency of cells capable of proliferation and multipotential differentiation in CD34⁺ CD38⁻ and CD34⁺thy-1⁺ population is very similar (1 in 182 in the CD34⁺ CD38⁻ population and 1 in 200 in the CD34⁺ thy-1⁺ population). These data also show that this ex vivo culture system is capable of measuring the 2 key stem cell–specific features, proliferation and multipotential differentiation, and suggest that this ex vivo culture system might represent a relevant system for the identification of primitive stem cells and for the relative position of different human HSC candidates in the stem cell hierarchy. Most importantly, this study shows that cells with the CD34⁺ CD38⁻ phenotype are expanded 1,500-fold in the mixed lymphoid/myeloid wells (Table 7). The frequency of reconstituted animals and the percentage of donor hematopoietic cells in the reconstituted animals derived from ex vivo–expanded CD34⁺ CD38⁻ cells purified from mixed lymphoid/myeloid wells are apparently very similar to the same parameters in the animals reconstituted with freshly purified CD34⁺ CD38⁻ cells from the same donors (Table 8). Based on the similar frequency of their activity to reconstitute the SCID-hu mice among freshly purified CD34⁺thy-1⁺ cells and CD34⁺ CD38⁻cells, and ex vivo–expanded CD34⁺ CD38⁻cells from mixed lymphoid/myeloid wells, we conclude that transplantable CD34⁺ CD38⁻ cells have been similarly expanded 1,500-fold as the cells with CD34⁺CD38⁻ phenotype in these mixed lymphoid/myeloid wells (Table 8). The bulk equivalent (ie, the overall culture, not just the 10% to 12% of mixed lymphoid/myeloid wells) is in excess of a 150-fold expansion. These results suggest that ex vivo expansion of transplantable human stem cells facilitated by this in vitro culture system is a general phenomenon and not just specific for CD34⁺ thy-1⁺ cells.

In this study, we have identified LIF as the determining cytokine in maintaining and expanding transplantable human CD34⁺thy-1⁺ cells in this ex vivo culture system. In addition, our studies demonstrate that LIF exerts its role in expanding transplantable human HSCs by indirectly affecting the stromal AC6.21 cells. Recently, it has been reported that addition of LIF to SyS-1 stromal cells enables the maintenance and 9-fold expansion of highly enriched competitive repopulating murine stem cells.31 The reverse transcription-polymerase chain reaction (RT-PCR) was used to show that M-CSF, IL-7, SCF, flt3/flk2 receptor, IL-2, IL-6, G-CSF, and LIF were upregulated on SyS-1 stromal cells upon LIF stimulation.31 Evidence was presented to suggest that synergy between IL-6 and SCF, both of which are upregulated by LIF on SyS-1 stroma, most likely accounts for the LIF-regulated murine stem cell maintenance and expansion in vitro.31 However, our data show that AC6.21 stroma treated with a combination of IL-3, IL-6, GM-CSF, and SCF is not capable of promoting ex vivo maintenance and expansion of CD34⁺ thy-1⁺ cells (Table 5). These results suggest that the synergy between IL-6 and SCF cannot account for the 1,500-fold expansion of human CD34⁺thy-1⁺ cells in our ex vivo expansion system. We hypothesize that LIF binds to the receptor on AC6.21 cells and initiates signal transduction pathways leading to upregulated production of factor(s) that facilitate transplantable stem cell expansion and/or downregulated production of factor(s) that inhibit stem cell expansion. Although the molecular mechanisms of LIF-mediated ex vivo maintenance and expansion of transplantable human fetal BM CD34⁺ thy-1⁺ and CD34⁺CD38⁻ cells are not currently defined, it is likely that this ex vivo culture system will provide a relevant tool for the identification of novel factors involved in regulating the process of self-renewal, proliferation, and differentiation in early hematopoietic development, and will have important implications for ex vivo stem cell expansion, gene therapy, and therapeutic transplantation. In this study, human fetal BM has been used as the HSC source. However, fetal BM is not a relevant HSC source in clinical application. The traditional sources of HSCs in clinical transplantation include autologous and allogeneic adult BM and mobilized peripheral blood (PB). Recently, human umbilical cord blood (UCB) has been shown to be an alternative source of stem cells for clinical transplantation.43 To evaluate the feasibility of using this ex vivo culture system to expand transplantable CD34⁺thy-1⁺ or CD34⁺ CD38⁻ cells for clinical transplantation, the ex vivo expansion potential of purified CD34⁺ thy-1⁺ or CD34⁺CD38⁻ cells from adult BM, adult PB, and UCB under this culture system should be determined.

We thank Drs Don Diamond, John Zaia, John Rossi, David DiGiusto, and Catherine M. Verfaillie for review of the manuscript; Drs Jeffrey Longmate and Joycelynne Palmer for their assistance in statistical analysis; Lucy Brown and Jim Bolen for assistance in FACS analysis; Drs Steve Novak and Tom LeBon for their assistance in the preparation of the manuscript; supportive team members in the Department of Molecular Biology at Beckman Research Institute at City of Hope for their administrative assistance; and members in the Animal Research Center at City of Hope for their assistance in animal care.