Lentiviral gene transfer and ex vivo expansion of human primitive stem cells capable of primary, secondary, and tertiary multilineage repopulation in NOD/SCID mice

Genetic modification of human hematopoietic stem cells (HSCs) is an appealing approach for the correction of many inherited or acquired defects of the hematopoietic system. To this end, the relevant gene must be delivered, integrated, and stably expressed in cells able to both self-renew and differentiate in all of the hematopoietic lineages. Only HSCs possess these properties. Several in vitro assays have been developed to detect pluripotent human hematopoietic cells,1-3 but they do not correctly reflect stem cell activity.4-6 The transplantation assay available in the mouse has been instrumental in defining and characterizing the most primitive cells of the hemopoietic system. A similar assay has been reported for human cells that measures the ability of human HSCs to reconstitute human hematopoiesis on a long-term basis in sublethally irradiated nonobese diabetic/severe combined immunodeficient (NOD/SCID) mice.7-10 

Recently, lentiviral vectors based on the human immunodeficiency virus type 1 (HIV-1) have been developed and shown to be efficient for the delivery and expression of genes in dividing and nondividing cells.11-13 As they do not require cell division for stable integration into the host genome, lentiviral vectors have proved to be most valuable for transduction of primitive HSCs, which usually reside in a quiescent status.14 Recent reports indicate that human cord blood (CB) CD34⁺ cells, possessing long-term repopulation ability in NOD/SCID mice, can be transduced with a lentiviral vector under conditions that are insufficient for transduction with retroviral vectors.14,15 Efficient transduction of primitive repopulating stem cells has also been recently demonstrated by secondary transplantation of NOD/SCID mice with lentivirus-transfected CB CD34⁺cells.16 

CB is an established source of HSCs for allogeneic or autologous transplantation,17 as in the case of children with adenosine deaminase (ADA) deficiency.18 As the limitations and risks associated with immunological barriers in acceptance of allogeneic grafts might be overcome by gene therapy with autologous genetically corrected stem cells,19 CB is regarded as a valid alternative to adult stem cell sources. The major limitation in widespread use of CB stem cells is its physiological small volume. Therefore, an insufficient number of genetically modified transplanted HSCs could result in an inability to compete with and eventually replace the defective host hemopoiesis. Thus, expansion of transplantable stem cells in vitro might prove extremely useful. However, the vast majority of studies reported so far with lentiviral vectors (LVs) have been aimed at developing gene transfer protocols that would allow maximum gene transfer with the least ex vivo culture to reduce negative effects on the survival and perhaps the long-term in vivo repopulation ability of stem cells.14-16 Only a few studies have reported suitable culture conditions for expansion of transplantable human CB stem cells.5,20-22 Among them, our previous studies, confirmed by others, have demonstrated that CB long-term repopulating stem cells can be expanded ex vivo for up to 10 weeks in the presence of flt 3/flk2 ligand (FL), thrombopoietin (TPO), stem cell factor (SCF), and interleukin-6 (IL-6).23-25 In addition, as recent studies have also demonstrated that CD34⁺ cells that are in G₁ or G₂/M are more effectively transduced than those in G₀,26,27 the cytokine requirements and cell cycle status of HSCs for lentiviral transduction remain controversial.28 

In the present study we show that advanced-generation LVs can be used to efficiently and durably deliver the green fluorescent protein (GFP) reporter gene into primitive CB HSCs, which are subsequently subjected to extensive ex vivo cultures. We demonstrate extended and abundant in vitro production of cells and progenitors expressing the marker gene and in vivo hemopoietic reconstitution by GFP-expressing myeloid and lymphoid progeny for 3 sequential generations of mice. These data indicate that ex vivo manipulation of human primitive HSCs does not necessarily lead to loss of self-renewal and multilineage differentiation capacity. Rather, gene transfer into primitive HSCs, followed by their ex vivo expansion, could provide means to facilitate and accelerate engraftment of genetically modified stem cells.

Umbilical CB was obtained, after written informed consent, at the end of full-term pregnancies, after clumping and cutting of the cord, by drainage of blood into sterile collection tubes containing the anticoagulant citrate-phosphate dextrose. CD34⁺ cells were obtained with MiniMACS separations as previously described.23 

The following recombinant purified human cytokines were used: recombinant human (rh) stem cell factor (rhSCF) and thrombopoietin (rhTPO) were generous gifts from Kirin (Kirin Brewery, Tokyo, Japan); interleukin 3 (rhIL3) was from Sandoz (Basel, Switzerland); erythropoietin (rhEPO; EPREX) was from Cilag (Milan, Italy); FLT3-ligand (rhFL) was kindly provided by S. D. Lyman (Immunex, Seattle, WA); rhIL6 was purchased from PeproTech (Rocky Hill, NJ).

Replication-defective self-inactivating HIV-1 vectors were constructed as described.13,29 Vectors were produced by transient-3 plasmid transfection of 293T. The 3 plasmids used were the transfer vector pRRLsin.PPT.hPGK.eGFP.Wpre,13 the VSV-G envelope-encoding plasmid pMD.G,11 and the packaging plasmid CMVΔR8.74.29 Viral supernatants were concentrated by ultracentrifugation. Titers of vector preparations were determined by transduction of HeLa cells with serial dilutions of vector supernatants, followed by cytometric analysis 3 days later. Final vector titers were in the range of 10⁹ transducing units (TU) per milliliter.

To induce cell cycling prior to transduction, 1 × 10⁵ purified CD34⁺ cells per milliliter were prestimulated for up to 96 hours with FL, SCF (both at 50 ng/mL), TPO (20 ng/mL), and IL6 (10 ng/mL). Infection of cells was carried out in flat-bottomed 24-well plates (Corning Costar, Cambridge, MA). For transduction, 10⁵ prestimulated CD34⁺cells were resuspended in 100 μL Iscove Modified Dulbecco Medium (IMDM; Gibco, Grand Island, NY) with 10% fetal calf serum (FCS; HyClone, Logan, UT). The cells were transduced with LVs at multiplicities of infection (MOIs) of 15 and 50, corresponding to the transducing units of 15 × 10⁶/mL and 50 × 10⁶/mL, respectively, for 12 hours in the presence of the same cytokine combination and incubated at 37°C and 5% CO₂. Then the cells were harvested, washed twice, and used to initiate both stroma-free expansion cultures and semisolid cultures (see “Cell culture assays”).

Assays for granulopoietic, erythroid, megakaryocytic, and multilineage granulocyte-erythroid-macrophage-megacaryocyte colony-forming units (CFU-GM, BFU-E, CFU-Mk, and CFU-GEMM, respectively) were performed as previously described.21,30,31 Hematopoietic colonies were scored after 14 days of culture and GFP⁺(fluorescent) colonies were identified by fluorescence microscopy.

Stroma-free expansion cultures for extended periods of time were performed in 24-well plates as previously reported.21,30Briefly, CB CD34⁺ cells (0.3-2 × 10⁴), unmanipulated or after transduction, were cultured in quadruplicate flat-bottomed 24-well plates in 1 mL IMDM + 10% FCS with FL, TPO, SCF, and IL6. Every week, all wells, after vigorous pipetting, were demidepopulated by removing one half of the cell suspension, which was replaced with fresh medium and growth factors. Harvested cells were used to assay colony-forming cell (CFC) content and for immunophenotype (CD34⁺, CD34⁺/GFP⁺cells). Expansion cultures for mouse transplantations were performed as reported.23 Briefly, 5 × 10⁴CD34⁺ CB cells/mL (unmanipulated or after transduction), resuspended in the same medium plus growth factors described above, were seeded in tissue culture T₂₅ flasks. Every week new culture medium was added, according to cell counts, and cells were transferred to additional flasks when necessary. At the start of cultures, at week 1, and at week 4, cell counts, immunophenotype (CD34⁺, CD34⁺/GFP⁺), and CFC content of the cells harvested from the expansion flasks was assessed.

After purification, aliquots of CD34⁺ CB cells were incubated with mouse immunoglobulin G (IgG; Fluka, Gallarate, Milano, Italy) and then stained for 30 minutes with anti–CD34-PE (Becton Dickinson, San Jose, CA) and anti-CD38–fluorescein isothiocyanate (FITC) (Dako A/S, Glostrup, Denmark) monoclonal antibody (MoAb) or the corresponding control antibodies in fluorescence-activated cell sorter (FACS) buffer containing phosphate-buffered saline (PBS), 0.1% bovine serum albumin (BSA), and 0.01% sodium azide. Cells were then washed, fixed for 10 minutes with 2% paraformaldehyde (Sigma, Gallarate, Milano, Italy), and resuspended in FACS buffer. After transduction of CD34⁺ progenitor cells with LVs and then once a week, aliquots of cultured cells were washed and then subjected to the same procedure to evaluate CD34 and GFP expression.

Flow cytometric analysis was performed with a FACSVantage SE (Becton Dickinson). At least 10 000 events were acquired for each analysis. Analysis was performed with CellQuest software (Becton Dickinson).

NOD/LtSz scid/scid (NOD/SCID) mice were obtained from Jackson Laboratories (Bar Harbor, ME) and maintained in the animal facilities of Centro di Immunogenetica ed Oncologia Specimentale (CIOS; Torino, Italy). Mice that were to receive transplants were irradiated at 6 to 8 weeks of age with 350 cGy total body irradiation from a¹³⁷Cs source and, 24 hours later, given a single intravenous injection of human CD34⁺ CB cells harvested from expansion cultures or after transduction, as described. When low numbers of unmanipulated CD34⁺ cells were to be transplanted, 8 × 10⁵ irradiated CD34⁻cells were coinjected as carrier cells. The same number of irradiated CD34⁻ cells were also injected as a negative control. Mice were killed 6 to 8 weeks after transplantation for assessment of number and types of human cells detectable in femurs and tibias.

BM cells were flushed from the femurs and tibias with a syringe and 26-gauge needle. The cells were resuspended at 1 to 2 × 10⁶ cells/mL and incubated with mouse IgG. Cells were then incubated with phycoerythrin (PE)–labeled MoAbs specific for human CD45 and glycophorin-A (GpA; Dako). Additional aliquots of cells were stained with antihuman CD14-PE (Dako), CD19-PE (Dako), CD41-PE (Dako), and CD34-PE (Becton Dickinson) in combination with antihuman CD45–tri-color (TC) (Caltag Laboratories, Burlingame, CA) to allow discrimination of subpopulations within the CD45 gate. Some cells from each suspension were similarly incubated with irrelevant (control) FITC-, PE- and TC-labeled MoAbs. After staining, all cells were washed once in FACS buffer. Then, contaminating red blood cells were lysed with EDTA (ethylenediaminetetraacetic acid) 10⁻⁴ M, KHCO₃ 10⁻³ M, NH₄Cl 0.17 M.

Gene transfer into different human progenitor (CD45⁺/CD34⁺), myeloid (CD45⁺/CD14⁺), B lymphoid (CD45⁺/CD19⁺), and megakaryocyte (CD45⁺/CD41⁺) subpopulations was determined by flow cytometric detection of GFP expression.

High-molecular-weight DNA was extracted from the BM of mice that had received transplants by phenol-chloroform extraction using standard protocols or by Blood and Cell Culture Kit (Qiagen Spa, Milan, Italy). DNA was digested with EcoRI (Gibco BRL, Grand Island, NY) and separated by agarose gel electrophoresis, transferred onto a positively charged nylon membrane, and probed with a labeled human chromosome 17-specific-α satellite probe (p17H8; limit of detection, approximately 0.05% human DNA). To quantify the level of human cell engraftment, the intensity of the characteristic 2.7-kb band in samples was compared with those of human:mouse DNA control mixtures (0%, 1%, 5%, 10%, 20%, 50% human DNA).

Quantitation of GFP DNA was obtained by real-time quantitative polymerase chain reaction (RTQ-PCR). PCR was performed in a final volume of 25 μL master mix containing 10 mM Tris (tris(hydroxymethyl)aminomethane)–HCl, pH 8.3, 50 mM KCl, 1.5 mM MgCl₂, dNTPs at 0.2 mM; forward and reverse primers at 15 pmol; and Taq polymerase (Eppendorf, Hamburg, Germany) at 5 U/100 μL. The GFP gene was amplified with GFP forward primer 5′-TGAAAGCGAAAGGGAA-3′ and GFP reverse primer 5′-GACTTCGCGCGTGCC-3′; the probe sequence is 5′-AGCTCTCTCGACGCAGGACTCGGC-3′. Probes and primers were designed using the Oligo 4.0 software (PrimerExpress Software Program), following guidelines suggested in the Model 7700 Sequence Detector instrument manual (PE Applied Biosystems, Foster City, CA). The exonuclease probe for the GFP was 5′ labeled with the reporter fluorescent dye VIC (5-fluorescein) and a fluorescence dye quencher, TAMRA (6-carboxy-tetramethyl-rhodamine), at the 3′ end, hybridized to the target.

All PCRs were performed in triplicate in a 96-well microtiter plate format on an ABI PRISM 7700 Sequence Detector System (PE Applied Biosystems). The thermal cycling conditions included 2 minutes at 50°C and 10 minutes at 95°C. Thermal cycling proceeded with 40 cycles at 95°C for 0.5 minutes and 60°C for 2 minutes.

The fluorescence increase of VIC and TAMRA was automatically measured during PCR. During the extension phase of the PCR, the nucleolytic activity of the Taq DNA polymerase cleaves the probe from the target and releases the reporter fluorescent dye from the vicinity of the fluorescence dye quencher. This process results in augmentation of a specific VIC fluorescence signal. For each sample, the ABI PRISM 7700 software provided an amplification curve constructed by relating the fluorescence signal intensity (ΔRn) to the cycle number. Cycle threshold (Ct) was defined as the cycle number at which the fluorescence signal was more than 10 SD of the mean background noise collected from the 3rd to the 15th cycle.

All reactions were performed in the Model 7700 Sequence Detector, which contains a Gene-Amp PCR System 9600. Reaction conditions were programmed on a Power Macintosh 7100 (Apple Computer, Santa Clara, CA) linked directly to the Model 7700 Sequence Detector. Analyses of data were also performed on the Macintosh computer. Collection and analysis software was developed at PE Applied Biosystems.

Genomic DNA (gDNA) samples extracted from the BM of engrafted mice were cut using EcoRI (Gibco BRL) to release fragments from the vector DNA and BamHI (Gibco BRL) to release fragments from the vector long-term repeats (LTRs) and the flanking gDNA. A 10-μg portion of each sample was separated on a 0.8% agarose gel by electrophoresis, denatured, and blotted over a nylon membrane. Blots were probed overnight in ULTRAhyb buffer (Ambion, Austin, TX) using a ³²P-labeled (Random Priming, Amersham Pharmacia Biotech, Littlechalfont, England) -1.3 kb fragment derived from the EcoRI-BamHI digestion. Oligoclonal populations of GFP-transduced HeLa cells were obtained by culturing them in limiting dilution conditions and their gDNA was used as a positive control (50% mixed with murine gDNA).

gDNA (500 ng) extracted from BM cells and HeLa GFP⁺ cells was digested with 10 U of XhoI (Gibco BRL) at 37°C for 2 hours. After enzyme inactivation (80°C for 20 minutes), 100 ng digested gDNA was ligated with 160 U of T4 DNA Ligase (New England BioLabs, Beverly) at 16°C for 12 hours. The PCR was done on 30 ng of ligated gDNA in a final volume of 50 μL that contained 2.5 U Taq polymerase (Master Taq Kit, Eppendorf), dNTPs (50 mM of each), MgCl₂ (1.25 mM), and primers (sense, 25 pmol: 5′-AGGCCCGAAGGAATAGAAGA-3′ and antisense, 50 pmol: 5′-CTGCTAGAGATTTTCCACACTGAC-3′). The conditions used were 95°C for 5 minutes; 35 cycles of 95°C for 1 minute, 58°C for 2 minutes, and 72°C for 3 minutes; and extension at 72°C for 10 minutes.

After electrophoresis on agarose gel and transfer to nylon membrane, PCR products were hybridized with the Rapid-hyb Buffer (Amersham Pharmacia Biotech) with a ³²P-labeled probe: 5′-GGTACAGTGCAGGGGAAAGA-3′. Oligonucleotides (5 pmol) were labeled with [³²P] adenosine 5′-triphosphate (ATP) 5000 Ci/mn (185 Bq/mn) (Amersham Pharmacia BIOTECH), using 20 U T4 polynucleotide kinase (New England BioLabs) at 37°C for 1 hour.

In these experiments CD34⁺ cells were first exposed to FL, SCF, TPO, and IL6 for 24 to 48 hours and then transduced with a vector concentration of 15 to 50 × 10⁶ TU/mL, corresponding to an MOI of 15 to 50, for 12 hours in the same medium. Aliquots of transduced cells were directly inoculated into sublethally irradiated NOD/SCID mice and engraftment by human cells was evaluated 6 to 8 weeks later. Levels of human engraftment were dependent upon the number of injected CD34⁺ cells, while the percentage of GFP⁺ cells increased slightly with the higher vector dose (Table 1). Multilineage engraftment was demonstrated by the presence of human cells belonging to granulocyte-macrophage, erythroid, megakaryocyte, and lymphoid lineages; GFP⁺ cells were found in all lineages. GFP⁺/CD34⁺ cells were present in all animals, indicating that transduced progenitor cells were at least maintained in the BM of mice that had received transplants.

To assess whether transduced cells retained their full capacity to undergo a long-lasting in vitro expansion, similar to that reported for unmanipulated CB CD34⁺ cells, we seeded small aliquots of both transduced and mock-transduced cells into 24-well plates in stroma-free suspension cultures in the presence of FL, TPO, SCF, and IL6 and subjected them to extended culture as described.21Both groups of treated cells behaved similarly: cells grew continuously for up to 30 weeks, reaching several million times the initial number (Figure 1). CD34⁺ cells underwent a parallel expansion, being still present after 24 and 28 weeks of culture in wells initiated with both control and transduced cells.

CFU-GM, BFU-E, and CFU-Mk were detected for up to 28 weeks of culture. During the entire culture period the presence of fluorescent cells and colonies was monitored. GFP-expressing CD34⁺cells were detected for up to 20 weeks of culture. At that time 18% of all cells were GFP⁺, 10.5% of the cells were CD34⁺, and 25% of CD34⁺ cells were GFP⁺. Committed progenitors were abundantly produced by the expanded cells. At week 18 the number of fluorescent colonies was more than 16 000 times the number of fluorescent colonies detected soon after transduction.

The in vitro long-term expansion data suggested that CB CD34⁺ cells, after lentiviral transduction, could still undergo a long-lasting expansion. As committed progenitors are short-lived, continuous production for longer than 4 months of GFP⁺ progenitors and cells belonging to all lineages indicated that some primitive hematopoietic cell(s) had integrated the transgene, which was capable of expression for at least 20 weeks.

The data generated by the above experiments prompted us to switch to larger-scale expansion experiments to verify whether among long-term expanded cells, putative stem cells had integrated the LVs and expressed the transgene. The most reliable assay to assess the presence of in vivo repopulating stem cells was the xenogeneic transplant in the NOD/SCID mouse: thus aliquots of transduced cells and their corresponding progeny collected after 1 and 4 weeks of expansion were injected into sublethally irradiated mice. The total number of cells and the number of CD34⁺ cells, after a 4-week expansion, were 213 ± 87 times and 43 ± 13 times the input, respectively (Table 2). These numbers are very similar to those obtained in demidepopulation wells (reported in Figure 1). Table 2 shows that in mice receiving similar numbers of expanded and nonexpanded CD34⁺ cells, the levels of human engraftment were higher in mice receiving transplants of the expanded cells, indicating that the transduction procedure did not impair the repopulating potential of CB CD34⁺ cells. The percentage of GFP⁺ human cells in mice injected with expanded cells was similar to that observed in mice that received transplants directly after transduction, suggesting that transgene integration and expression were not hampered by the ex vivo expansion. Also, all animals that received transplants were multilineage engrafted. Each subpopulation contained similar proportions of GFP⁺ cells, and 15% to 44% of CD34⁺ cells were GFP⁺, indicating that the expansion procedure at least was not detrimental to the maintenance or expansion of the transduced CD34⁺population (Figure 2).

To assess clonal contributions to the engraftment, we performed Southern blot analysis on the GFP⁺ BM samples from 2 mice, each injected with 2 × 10⁵ transduced CB CD34⁺ cells that had been expanded for 4 weeks with FL, TPO, SCF, and IL6. The BM DNA was digested with either double- or single-restriction enzyme cutting the internal region of vector LTRs to reveal all transduced cells or single clonal populations, respectively. In our samples, when the same GFP probe was used for both DNA-released fragments (EcoRI fragment and BamHI fragment), only the population of all transduced cells could be detected; no distinct bands were seen on BamHI-digested samples (Figure3), suggesting the presence of multiple integrants below the detection limit.

To better assess the number of clones involved, we performed an inverse PCR on the BM samples32; at least 4 different integrants in BM A and 2 in BM B could be detected (Figure 3), indicating that a limited, but well-represented, number of cell populations were present. This result suggests that probably few GFP⁺ clones, after 4 weeks of expansion in our culture conditions, can engraft and repopulate the BM of mice that have received transplants.

Serial transplantations were performed to prove that very primitive stem cells had indeed been hit in the transduction procedure, survived the 4-week in vitro manipulation, and did not lose their self-renewal ability and multilineage differentiation capacity. BM cells from 5 primary mice were injected into 12 secondary sublethally irradiated mice (Table 3). No growth factors were administered to the animals. Each mouse was injected with unfractionated BM (2-10 × 10⁶, except for 3 mice that received 2 × 10⁵ human CD45⁺-enriched cells). Human engraftment was assessed at weeks 6 to 8 after transplantation. Human CD45⁺ cells represented 13.17% ± 5.7% of the entire BM; human CD45⁺ cells expressing GFP⁺ represented 16.7% ± 7.5%. Interestingly, within the CD45 population, CD34⁺ cells constituted a good percentage (9.8%-25%), and 25% to 35% were GFP⁺. Multilineage engraftment was consistently and reproducibly obtained in all cases. GFP⁺ cells were found in all different lineages (Figure 2). Upon plating in plasma clot cultures, fluorescent erythroid and myeloid colonies were detected (data not shown).

In 6 separate experiments, tertiary serial transplantations were performed. Total BM cells (8-20 × 10⁶ per mouse) were inoculated in these cases. Levels of engraftment were 5.9% ± 2.2%. CD45⁺ cells expressing GFP were 13.6% ± 4.2%. Fluorescent myeloid and erythroid colonies were observed when BM cells were cultured in semisolid assays (not shown). All transplant recipients were multilineage engrafted; all cell subsets contained GFP⁺ human cells (Figure 4). Primary, secondary, and tertiary engraftment by GFP⁺ cells was confirmed by Southern blot and by real-time PCR (Table 3, Figures 4and 5). In particular, Figures 4 and 5directly compare the data generated by FACS analysis, Southern blot, and quantitative real-time PCR on the same animals. These data suggest on average a single integrant per cell (as the copy number is similar to the percentage of positive cells by FACS).

Interestingly, the percentage of GFP⁺ cells remained similar in the serial transplants; this may suggest that even at the third transplant, limiting numbers of SRCs were not injected. Table4 summarizes the results of an additional experiment in which initial CD34⁺ cells (3 × 10⁵), after 96 hours' pretreatment with growth factors and 24 hours' exposure to lentiviral particles, were further expanded for 4 weeks. At the end of culture, cells were collected and the expansion equivalent of 6 × 10⁴ initial cells were inoculated into each of 4 mice. Secondary mice received transplants of 75 × 10⁴ CD45⁺ cells isolated by the pooled BM of the 4 primary mice. Five tertiary mice received transplants of 12 × 10⁶ total BM cells of the pooled BM harvested from the 4 secondary mice (4 of 5 mice survived). The presence of human cells, some of which were GFP⁺, was also in this case confirmed by Southern blot analysis (Figure 5) and real-time PCR (Figure 6).

As previous limiting dilution studies showed that in such culture conditions SRCs underwent a net expansion (W.P. et al23and manuscript in preparation, 2002), the same conditions were employed, after transduction, to verify whether the transgene was integrated into the genome and was expressed by primitive stem cells that could self-renew and generate identical progeny that, in turn, retained identical self-renewal and multilineage differentiation abilities. CD34⁺ cells were subjected to gene transfer: following transduction, an aliquot of 200 000 CD34⁺ cells was cultured ex vivo for a total of 4 weeks. All cells were collected and inoculated at decreasing doses into NOD/SCID mice. Four mice, each injected with the expansion equivalent of 30 000 CD34⁺cells, were killed 6 weeks after transplantation. All mice were engrafted (Table 5). Levels of human CD45⁺ cells were 9.6% ± 2%. Multilineage engraftment was clearly detectable. GFP⁺ cells represented 2.5% of the entire BM; CD45⁺ cells expressing GFP constituted 17.9% ± 3.2%. GFP⁺ cells were present in all lineages. All mice injected with the expansion equivalent of 20 000 CD34⁺ cells were engrafted (CD45⁺ level was 4.9% ± 1.1%; GFP-expressing CD45⁺ cells were 14.2% ± 4.6%). Secondary transplantations (performed with the BM cells of primary recipients that had received the expansion equivalent of 30 000 and 20 000 CD34⁺ cells) showed that human engraftment was achieved in all mice, indicating that 30 000 and 20 000 initial CD34⁺ cells contained at least one SRC that had stably integrated into the transgene and that the progeny of that SRC was at least a second SRC, which still integrated the transgene. In all cases the transgene was integrated and capable of expression in all hemopoietic lineages. Four additional mice were injected with the expansion equivalent of 10 000 CD34⁺ cells, and all were found positive for human engraftment. Inasmuch as good levels of engraftment were detected in all animals with 10 000 expanded CD34⁺ cells, we may hypothesize that the expansion equivalent of 10 000 initial cells is not a limiting dilution and that in this population more than 1 SRC is likely to exist. As the engrafted cells did not contain only GFP⁺ or GFP⁻ cells, at least on FACS analysis, we might conclude that at least 2 SRCs were present before the gene transfer procedure and that at least one of these integrated the transgene. During the ex vivo culture both transduced and nontransduced SRCs were maintained or even expanded.

Taken together, these data show that after transduction both CD34⁺ and CD34⁺/GFP⁺ cells can be expanded in vitro for 4 weeks, after which time they maintain long-term repopulating capacity for at least 3 generations, which implies a very large production of cells and GFP-expressing cells in vivo.

The studies reported here show that primitive long-term repopulating HSCs can be effectively transduced by advanced-generation LVs, yield large outputs of committed progenitors, and do not lose their proliferation potential, self-renewal ability, multilineage differentiation ability, and capacity for stable transgene expression.

Experimental results are provided from 2 different sources: (1) an in vitro assay of extended long-term cultures and (2) in vivo long-term repopulating assays, performed as serial transplantations in the xenogeneic NOD/SCID mouse model. The gene transfer procedure was carried out with a 1- to 4-day preincubation period in serum-repleted, stroma-free liquid cultures in the presence of a growth factor combination shown to sustain maintenance or even expansion of human HSCs.21,24,25 

Although LVs are well known for their ability to effectively transduce nondividing cells as well, and in particular human HSCs,11-13 we decided to employ growth factor prestimulation in an attempt to show that ex vivo manipulation can be associated with lentiviral gene transfer to improve the engraftment ability of genetically modified stem cells.

Soon after transduction cells were grown in vitro for long periods of time in culture conditions that have been previously described.21 Significant differences in the proliferation potential (number of cells generated over time) and committed progenitor output between transduced and mock-transduced cells could not be identified. Thus it may be supposed that the transduction procedure and the culture conditions described here do not affect these properties of more primitive hemopoietic progenitors.

However, initial integration of the transgene into HSCs does not guarantee transgene expression in their progeny. Under the culture conditions reported here, transgene expression could be easily detected for more than 20 weeks: at that time the percentage of GFP⁺cells was quite high in both total and CD34⁺ cells. Also, GFP⁺ colonies were still detectable after 18 weeks of culture. The continuous, increasing production of cells and of committed progenitors over such a long time period is a clear suggestion that some primitive HSCs had successfully integrated the transgene and that its expression was maintained during differentiation. This finding is in agreement with previous reports.14,15,28 However, the gene transfer conditions in these studies were different from ours: the culture conditions minimized stem cell cycling by shortening preincubation of the cells and eliminating or reducing cytokines and serum from the transduction preparation. Loss of in vivo repopulating activity by cytokine-prestimulated, retroviral-transduced CB cells has also been reported, but again, culture conditions and cytokines used were different from those adopted in the present study.33 

Lentiviral-mediated gene transduction and expression in in vivo experiments are particularly important. Indeed, the development of the NOD/SCID xenotransplantation system has enabled the detailed characterization of primitive HSCS, capable of long-lasting and complete hemopoietic reconstitution.7-10 

In the experiments reported here transduced cells were expanded ex vivo for up to 4 weeks after transduction. Mice inoculated with expanded cells showed higher levels of human engraftment than mice injected soon after gene transfer. More important, the percentage of GFP⁺cells was similar to that obtained in mice that received transplants of nonexpanded CB cells. These data indicate that the engraftment ability of transduced SRCs was not impaired but rather was maintained or increased by the ex vivo expansion. Immunophenotyping of the engrafted human cells showed evidence of multilineage engraftment. The levels of GFP-expressing cells were also quite similar in all subpopulations, indicating that a primitive SRC with both lymphoid and myeloid differentiation potential had been successfully transduced and maintained, if not expanded, during the 4-week ex vivo culture.

Secondary and tertiary serial transplantations showed good levels of human multilineage engraftment in all recipients. A portion of human cells were GFP⁺, indicating that very primitive SRCs had been transduced and that good levels of the transgene expression could be achieved for at least 3 generations of mice. Southern blot and real-time PCR data confirm these data and also indicate the stable integration of the transgene, which is capable of expression (as it is FACS detectable). In other words, transgene down-regulation is not occurring (a phenomenon we did observe when MoML-V–based vectors were used [W.P. et al, manuscript in preparation, 2002]). Successful secondary transplantations have been reported so far only by Woods et al, using primary, unmanipulated CB cells that were transduced overnight by LVs.16 In our experiments transduced CB cells were expanded ex vivo for 4 weeks prior to injection into NOD/SCID mice, which suggests that the GFP transgene was successfully delivered in one or more SRCs that retained extensive self-renewal capacity and were able both to fully reconstitute the hemopoiesis in secondary and tertiary recipients and to express the transgene. It was evident that the ex vivo expansion did not seem to impair the self-renewal potential or the transgene expression of primitive stem cells and their progeny. While these findings might rule out a carryover of more mature GFP⁺ hematopoietic progenitors, they do not completely rule out a carryover of deeply quiescent SRCs. Also, the percentage of GFP⁺ cells in primary, secondary, and tertiary mice remained constant, which might suggest that even in tertiary mice SRCs were not at limiting dilutions. Down-regulation of expression of a transferred gene in the case of retroviral delivery vectors is a phenomenon recognized as a threat to the efficacy of gene therapy. LVs enable stable long-term transgene expression in transduced cells, including the differentiated progeny of HSCs. Long-term transgene expression was observed in the tissues directly transduced in vivo.11 Most papers published today, including ours, provide evidence for stable transgene expression in human hematopoietic grafts of NOD/SCID mice receiving transplants of LV-transduced SRCs.13,15,28 ^34–36 Because of the prolonged culture time and the extensive cellular expansion obtained in our work, in vitro and in vivo, the observed stable transgene expression provides a new important validation of LVs.

Further experiments performed by inoculating NOD/SCID mice with decreasing numbers of expanded CB cells led us to detect successful multilineage engraftment in all mice that received transplants of the expansion equivalents of 30 000, 20 000, and 10 000 initial CD34⁺ cells. As 20 000 or more unmanipulated CB CD34⁺ cells represented a limiting dilution in other studies and in our laboratories,16,23,24 the present finding that the expansion equivalents of 10 000 CD34⁺cells abundantly engrafted the total BM of the mice that received transplants indicates that the SRC(s) present in the initial cell population underwent expansion in the following 4 weeks of ex vivo culture. Indeed, in previous work we and others showed that the frequency of SRCs in CB cells expanded for 8 to 10 weeks is 1 in 350 or 375 initial CD34⁺ cells. Accordingly, the data provided from assessment of GFP expression in human CD45⁺ cells fail to show the exclusive presence of CD45⁺/GFP⁺ cells in mice that underwent transplantation, as has recently been reported by others.16 This may constitute further proof that we transplanted several SRCs, some of which had integrated and still expressed the transgene. Ongoing experiments performed by injecting mice with fewer expanded cells seem to confirm our hypothesis (W.P., manuscript in preparation, 2002).

To assess clonal contribution to the engraftment by transduced and expanded cells, we performed Southern blot and Inverse PCR studies. As we could detect a band on the EcoRI-digested samples, but not onBamHI-digested ones, we could argue that likely more than one integrant is contributing to the engraftment. Our inverse PCR data also show that more than one integrant is involved in the engraftment. Although these experiments are preliminary, taken together they support the hypothesis that more than one clone is contributing to the engraftment.

The present study demonstrates that use of LVs and culture conditions that are most suitable for the maintenance and even expansion of primitive hemopoietic repopulating stem cells results in high levels of engraftment by stem cells that closely resemble their nontransfected counterparts. In other words, in addition to retaining extensive self-renewal abilities and multilineage differentiation capacity, CB CD34⁺ cells and their progeny integrate and express the transgene for long periods of time in vivo.

Far from being definitive, these data may represent another step toward the identification of optimum conditions for the implementation of preclinical protocols.

The authors wish to thank Kirin and Immunex for continuous supply of growth factors.