Efficient c-kit Receptor-Targeted Gene Transfer to Primary Human CD34-Selected Hematopoietic Stem Cells

Abstract

We have previously reported effective gene transfer with a targeted molecular conjugate adenovirus vector through the c-kit receptor in hematopoietic progenitor cell lines. However, a c-kit-targeted recombinant retroviral vector failed to transduce cells, indicating the existence of significant differences for c-kit target gene transfer between these two viruses. Here we demonstrate that conjugation of an adenovirus to a c-kit-retargeted retrovirus vector enables retroviral transduction. This finding suggests the requirement of endosomalysis for successful c-kit-targeted gene transfer. Furthermore, we show efficient gene transfer to, and high transgene expression (66%) in, CD34-selected, c-kit⁺ human peripheral blood stem cells using a c-kit-targeted adenovirus vector. These findings may have important implications for future vector development in c-kit-targeted stem cell gene transfer.

A major goal of gene therapy is to develop vectors that stably transduce hematopoietic stem cells (HSC) (21, 32). Success with this strategy could potentially result in cures for genetic diseases such as immunodeficiency syndromes, cancer, storage diseases, or sickle cell disease (14, 23, 30, 45, 47). At present, low transduction efficiencies and loss of transgene expression are major obstacles to be overcome in order to develop hematopoietic stem cell gene therapy for clinical purposes (17, 20, 21, 47).

One reason for poor transduction efficiency with virus-derived vectors is low viral receptor expression on immature progenitor cells (11, 13). These receptors are responsible for the natural tropism of viruses to cells (3, 5, 31). The c-kit receptor is a cell surface marker which is coexpressed on immature CD34⁺ hematopoietic stem cells (8, 19, 33, 46, 50). Therefore, the c-kit receptor has been identified as a potential specific entry port for targeting stem cells. The restricted tropism of wild-type or pseudotyped vectors can be overcome by redirecting virus-derived vectors through c-kit. The feasibility of c-kit receptor-targeted gene transfer was first shown in hematopoietic progenitor cell lines with a transiently expressing adenovirus (Ad)-based molecular conjugate vector (43). In order to achieve persistent transgene expression in c-kit-positive cell lines, two other groups engineered recombinant retroviral vectors expressing the c-kit ligand stem cell factor (SCF) on the envelope. Although both groups demonstrated specific binding and vector uptake via c-kit compared to the control retrovirus vector, the c-kit-targeted vector failed to achieve cellular transduction as measured by reporter gene expression (15, 49). These conflicting data suggested that c-kit receptor-targeted gene transfer cannot be universally applied, since the system is vector dependent.

This work was designed to address two critical issues in c-kit-targeted gene transfer. First, we hypothesized that c-kit-mediated gene transfer would require efficient endosomalytic activity for successful transgene expression. Toward this end, we conducted experiments in MO7-e cells with a c-kit-retargeted retrovirus with and without endosomalysis. To demonstrate feasibility for c-kit-targeted gene transfer in CD34-selected human hematopoietic stem cells, proof-of-principle experiments with a recombinant adenovirus capable of efficient endosomalysis were conducted.

MATERIALS AND METHODS

Construction of Ad-EGFP.

Enhanced green fluorescence protein (EGFP) cDNA was released from plasmid pEGFP-N1 (Clontech, Palo Alto, Calif.) by digestion with the restriction enzymes EcoRI and XbaI (New England Biolabs, Beverly, Mass.). The gene was separated by gel electrophoresis, purified, and subcloned into the EcoRI and XbaI sites of pACCMV.PLA (24). This vector was cotransfected into 911 cells with XbaI-restricted AdCMVLacZ DNA (24) using calcium phosphate precipitation (18), and plaques were screened by blue-white selection as described by Schaack et al. (39). Ad-EGFP clones were screened by PCR, and protein production was confirmed by fluorescent microscopy of 911 cells. One of these clones was chosen for all subsequent experiments.

Viruses were propagated on 911 cells using endotoxin-free conditions and were purified by CsCl as previously described (24, 25). Viral preparations were screened for replication-competent adenovirus by propagation on A549 cells. This assay has a sensitivity of 1 contaminant per 10⁸ PFU. All viral preparations had a PFU/particle ratio of <100:1. All lots of recombinant adenovirus contained less than 1 endotoxin unit/ml as measured by the Limulus amebocyte lysate assay (BioWhittaker, Walkersville, Md.).

Biotinylation of SCF.

Recombinant human SCF (rhSCF) was generously provided by Amgen (Thousand Oaks, Calif.). Centricon centrifugal filtering devices (Amicon, Inc., Beverly, Mass.) with a molecular weight cutoff of 10,000 were used for buffer-exchanging rhSCF with NaHCO₃ buffer (0.1 M NaHCO_3, pH 8.4). The filter was preblocked with 0.1% bovine serum albumin (BSA) in phosphate-buffered saline (PBS). A total of 100 μg of rhSCF (final concentration, 1 μg of rhSCF/μl of PBS) was then added to the column and diluted with 1.5 ml of NaHCO₃ buffer. The NaHCO₃ buffer was removed by centrifugation at 4,800 × g at 4°C for 30 min. This procedure was repeated twice. The sample was then centrifuged at 4,800 × g for 1.5 h to achieve a final concentration of 1 μg of rhSCF/μl of NaHCO₃ buffer. One microliter of biotin-NHS (Calbiochem, La Jolla, Calif.) (10-μg/μl stock solution in dimethyl sulfoxide) was added, and the reaction mixture was incubated at room temperature for 30 min. This was followed by three buffer exchanges against PBS to remove free biotin. Ten-microgram aliquots were then lyophilized and cryopreserved at −20°C for further use. The bioactivity of biotinylated SCF (SCFbiot) was determined to be equivalent to 90% that of unbiotinylated SCF by use of a previously described bioassay with factor-dependent MO7-e cells that measures proliferation by tritiated-thymidine incorporation (7).

Retroviral preparations.

Cells of the amphotropic retroviral packaging cell line RetroPack PT67 and the ecotropic retroviral packaging cell line EcoPack-293 (Clontech) were transfected with plasmid pMSCV-EGFP using calcium phosphate and selection with G418 (0.5 mg/ml) for 2 weeks. The retroviral helper cell lines were cultured under standard conditions (28). Supernatants were collected from stable vector-producing cells, and retrovirus purification was performed by a procedure previously described by Akatsuka et al. (1). Briefly, a 30% (wt/wt) stock solution of polyethylene glycol 8000 (PEG 8000) (Sigma, St. Louis, Mo.) was prepared in double-distilled water and stored in aliquots at 4°C. Viral supernatants were gently mixed with PEG in 250-ml polystyrene tubes to achieve a final 8% PEG solution. The mixture was maintained overnight at 4°C. After centrifugation at 1,500 ×g for 45 min, the precipitate was dissolved in 3 ml of TES buffer (10 mM Tris-HCl [pH 7.2], 2 mM EDTA, 150 mM NaCl). Titers of virus preparations were determined as previously described on NIH 3T3 monolayers (10, 38).

Biotinylation of retrovirus.

Ecotropic and amphotropic retroviruses (ectotropic Moloney murine leukemia virus [eMMLV] and aMMLV, respectively) were covalently linked to sulfo-NHS-biotin (biotin) by following the instructions of the manufacturer (Pierce, Rockford, Ill.). Briefly, 0.2 ml of biotin (10 mg/ml in dimethyl sulfoxide) was mixed with 2.8 ml of retrovirus (10⁷ CFU/ml) in PBS. The reaction mixture was incubated on ice for 2 h. The products (eMMLV-B and aMMLV-B) were then TES buffer exchanged (0.01 M Tris-HCl [pH 7.2], 0.002 M EDTA, 0.15 M NaCl) over EP 10 DG Columns (Bio-Rad, Hercules, Calif.). eMMLV-B and aMMLV-B were aliquoted and stored in TES at −80°C. Virus titers were determined as previously described (27, 28).

Adenovirus-avidin linkage and Cy3 labeling.

Sepharose CL4B columns (Pharmacia Biotech, Piscataway, N.J.) were equilibrated with HEPES-buffered saline (HBS, consisting of 5 mM HEPES [pH 7.8] and 150 mM NaCl) and loaded with 4 ml of Ad-EGFP or AdCMVLuc (for adenovirus-retrovirus conjugate experiments) (titer, 2 × 10¹¹ PFU/ml). The virus was eluted with 2 ml of HBS, and the final volume was adjusted to 3.6 ml with HBS. Then 2.38 mg of neutravidin (Pierce) diluted in 0.4 ml of HBS and 40 μl of 0.13 M 1-ethyl-3-(3-dimethylaminopropyl) carbodimide HCl (EDC; Pierce) in HBS solution were added. The reaction mixture was incubated on ice for 4 h. The avidinylated adenovirus (Ad-EGFP-Av) was purified on a CsCl density gradient and cryopreserved in virus preservation medium as previously described (41). Plaque assays determined a 50% loss in infectious activity after the avidinylation procedure. Effective avidinylation of adenovirus was determined by enzyme-linked immunosorbent assay (ELISA). ELISA plates (Nalge Nunc, Intl., Naperville, Ill.) were coated with varying concentrations of either Ad-EGFP-Av or unmodified adenovirus (Ad-EGFP) in carbonate buffer, pH 9.5, at a total volume of 100 μl/well overnight. Plates were washed four times with wash buffer (PBS with 0.05% Tween 20) and then incubated with blocking buffer (200 μl of PBS/well plus 0.5% BSA, 0.05% azide, and 2% skim milk) for 1 h at room temperature. Wells were then washed four times with wash buffer and incubated with a primary polyclonal anti-avidin antibody (Sigma) at room temperature for 1 h. After three more washing steps with wash buffer, wells were incubated with conjugated goat anti-rabbit alkaline phosphatase (Bio-Rad) (diluted 1:1,000 in PBS with 0.5% BSA). One hundred microliters of a 1% substrate solution in diethanolamine buffer (Sigma) was added and incubated for 15 min, and plates were analyzed at 410 nm on a plate reader. Luciferase-encoding adenovirus (AdCMVLuc) (24) was avidinylated according to this procedure and quality controlled (AdCMVLuc-Av).

Cy3 labeling of Ad-EGFP or Ad-EGFP-Av was accomplished by utilizing the procedure of Leopold et al. (26). Briefly, 100 μl (3 × 10⁹ PFU) of Ad-EGFP or Ad-EGFP-Av was added to 900 μl of “labeling buffer” (0.1 M HCO₃ buffer, pH 9.3). The quantity of Cy3 (Amersham) prescribed by the manufacturer for labeling 1 mg of protein was diluted in 160 μl of labeling buffer, and 4 μl of this solution was added to the adenovirus. After 1 h of incubation at room temperature, the reaction mixture was dialyzed against exchange buffer overnight at 4°C using 8,000-molecular-weight-cutoff dialysis tubing (10% glycerol, 50 mM Tris-HCl [pH 7.5], 250 mM MgCl₂). The final product (Ad-Cy3 or Ad-Av-Cy3) was aliquoted and stored in viral preservation medium at −80°C.

Cell culture and cell isolation.

The factor-dependent megakaryoblastic progenitor cell line MO7-e (2) was grown in RPMI supplemented with 10% fetal bovine serum (FBS; Gibco, Gaithersburg, Md.) and 25 ng of granulocyte-macrophage colony-stimulating factor (GM-CSF; Immunex, Seattle, Wash.)/ml in a humidified incubator under a 5% CO₂ atmosphere. Peripheral blood stem cells were obtained after G-CSF mobilization from two normal donors. The study was approved by the institutional review board at the Louisiana State University Health Sciences Center (LSUHSC), New Orleans. The mononuclear fraction was obtained by cytopheresis and purified for CD34-expressing cells using SEPRATE SC (stem cell concentration system) (Cellpro Systems, Bothell, Wash.). Flow-cytometric analysis was performed for CD34 antigen expression using a directly conjugated phycoerythrin (PE) HPCA-II-CD34 monoclonal antibody (Becton Dickinson, San Jose, Calif.), a directly conjugated fluorescein isothiocyanate (FITC)-CD45 antibody (Beckman-Coulter, Hialeah, Fla.), and 7-amino-actinomycin-D (7-AAD; Molecular Probes, Eugene, Oreg.) for viability (FACScalibur Flow Cytometer; Becton Dickinson). Briefly, cells were pelleted, resuspended, labeled simultaneously by adding each reagent, and then analyzed by first gating on CD45⁺ cells versus side scatter. Gated CD45⁺ cells were then gated for CD34 antigen and gated for only bright CD34 cells. CD34⁺ cells were gated for complexity and CD34 antigen density. Viability testing was performed on all CD45⁺ cells versus 7-AAD negativity. CD34⁺ cell viability testing was also performed on the gated clustered cells versus 7-AAD negativity. The CD34⁺ cells were 82% pure and >80% viable. Staining for c-kit was done using a directly FITC labeled anti-CD117 monoclonal antibody (PharMingen), and staining for CD38 was done with a directly FITC labeled anti-CD38 monoclonal antibody. Sixty-nine percent of CD34⁺ cells also coexpressed c-kit. Cells were analyzed for coxsackie adenovirus receptor (CAR) expression using the anti-CAR immunoglobulin G1 (IgG1) monoclonal antibody RmcB (5), (kindly provided by Robert Finberg, Boston, Mass.) and a goat anti-murine biotinylated antibody (PharMingen) followed by streptavidin Cy-chrome staining (PharMingen). Nonspecific binding was blocked using mouse IgG. The human lung cancer cell line A549 was obtained from the American Type Culture Collection (Manassas, Va.) and cultured in Dulbecco's modified Eagle medium plus 10% fetal calf serum (FCS).

Cell transduction and flow-cytometric analysis.

All vector transfection steps were performed at 4°C, unless indicated otherwise. MO7-e cells (5 × 10⁶/sample) were washed twice with cell wash buffer (PBS plus 0.5% BSA) and incubated on ice with 100 ng of SCFbiot (or unbiotinylated SCF as indicated for a control) for 60 min in a total volume of 100 μl of wash buffer. Cells were then washed with cell wash buffer to remove excess unbound ligand. To form c-kit-targeted retrovirus vectors, cells were resuspended in 100 μl of cell wash buffer and incubated with neutravidin (2 μg/sample) for 30 min. Excess neutravidin was removed by washing with cell wash buffer. Biotinylated retrovirus (eMMLV-B or aMMLV-B) was then added for assembly of the complete c-kit-retargeted retrovirus construct. For assembly of the retargeted retrovirus-adenovirus conjugate, cells were labeled with SCFbiot and excess unbound ligand was removed with a washing step. AdCMVLuc-Av (24) was then added, and the cell-suspension was incubated for an additional 30 min, followed by a washing step to remove unbound adenovirus. The cells were then incubated with biotinylated retrovirus, which was expected to bind to available unoccupied avidin sites of the retargeted adenovirus conjugate (SCFbiot-AdCMVLuc-Av). Excess retrovirus was removed by washing after a 30-min incubation step.

Compared to that for MO7-e cells, a slightly modified transduction procedure was used for primary CD34-selected HSC. SCFbiot or SCF was resuspended at a final concentration of 100 ng/μl in cell wash buffer. Ten microliters of this stock solution was added to 5 × 10⁷ PFU of Ad-EGFP-Av or Ad-EGFP as indicated. This reaction mixture was incubated for 30 min at room temperature. The solution was then added to 1.4 × 10⁵ CD34-enriched human stem cells suspended in 200 μl of Iscove's medium supplemented with 2% heat-inactivated FBS (Gibco). Where indicated, CD34⁺ HSC were preincubated with the SR1 antibody at 2 μg/ml (generously provided by Virginia Broudy, Seattle, Wash.) or its isotype control (9, 34). MO7-e cells were resuspended in 200 μl of RPMI medium supplemented with 2% heat-inactivated FBS (Gibco) and 25 ng GM-CSF (Immunex)/ml and were incubated in a rotating, prewarmed hybridization oven at 37°C for 1.25 h.

After the transfection procedure, cells were placed in 6-well plates at a final volume of 3 ml of their standard culture media. Analysis for EGFP expression of MO7-e cells was performed by flow cytometry 96 h posttransduction (FACScalibur; Becton Dickinson). Human HSC were analyzed by flow cytometry for EGFP expression and CD34 expression at 40 h.

The human lung cancer cell line A549 was cultured in Dulbecco's modified Eagle medium with 10% FCS (Gibco). Transduction of A549 cells was performed in near-confluent monolayer cultures under serum-reduced conditions (2% FCS) for 2 h. Cells were analyzed for EGFP expression at 72 h by transmission light microscopy.

Confocal microscopy.

CD34-selected peripheral blood stem cells (PBSC) or A549 cells were transfected as outlined above using Ad-Cy3 or the c-kit-targeted Cy3-labeled vector (SCFbiot + Ad-Av-Cy3). Cells were rinsed and kept for 1 h at 4 or 37°C as indicated. Cells were then fixed in 1% formaldehyde overnight at 4°C. Nuclei were counterstained with 4′,6′-diamidino-2-phenylindole (DAPI; Molecular Probes). Cells were analyzed on a confocal microscope at 1-μm increments (Noran Odyssey, Middleton, Wis.). Images were generated and analyzed using Metaview software (Universal Imaging, West Chester, Pa.).

Statistical analysis.

Data were analyzed by analysis of variance using the statistical program StatView (Abacus Concepts, Calabasas, Calif.). A P value of <0.05 was considered statistically significant.

RESULTS AND DISCUSSION

Formation of aMMLV-adenovirus conjugates enhances retroviral transduction efficiency in MO7-e cells.

Compared to unmodified retrovirus transduction, previous reports on retrovirus receptor retargeting consistently showed a substantial reduction in (Epo receptor), or complete loss of (c-kit receptor), cellular transduction (15, 22, 49). In contrast, our group successfully used polylysine-based adenovirus molecular conjugate vectors and recombinant adenovirus to accomplish c-kit-mediated cellular transduction (43, 44). We hypothesized that the retroviral transduction efficiency of cells is dependent on the entry port for the vector, so that c-kit-redirected vectors lose the ability to transduce the host genome compared to unmodified viruses (15, 49). These findings suggest that under physiological conditions, retroviral cell entry via its natural receptor facilitates the virus's life cycle, whereas c-kit-redirected retroviral cell entry can result in its disruption. In contrast to adenovirus or adenovirus molecular conjugate vectors, a retrovirus by itself has no endosomalytic properties. We hypothesized that the alternative entry pathway through c-kit redirects the retrovirus to the endosomal/lysosomal compartment, resulting in virus inactivation (12, 48). This hypothesis was supported by the observation of Yajima and coworkers that chloroquine treatment to some degree restored cellular transduction for a c-kit-targeted retroviral vector (49). To demonstrate that endosomalysis is required for c-kit-mediated gene transfer, different retrovirus vectors (MMLV) encoding the EGFP reporter gene were synthesized.

The biotin-streptavidin technique was used to incorporate adenovirus for endosomalysis and the SCF moiety for targeting (Fig. 1). Formation of aMMLV-adenovirus conjugates enhanced transduction efficiency almost threefold (Fig. 2A). Physical linkage of both compounds was required for this enhancement effect. Physical incorporation of the targeting ligand SCF into the complex (SCFbiot) enhanced transduction efficiency by approximately 50% compared to the untargeted construct (Fig. 2A). These results demonstrate that introduction of endosomalysis enhances the transduction efficiency of a retrovirus vector. However, amphotropic retrovirus was used for these experiments, so it is possible that uptake occurred via both amphotropic and c-kit receptors. To exclude uptake via the amphotropic receptor, experiments were conducted with eMMLV.

FIG. 1.

c-kit-targeted gene transfer with conjugates. Human MO7-e cells express the c-kit receptor (natural receptor for SCF) and aMMLV receptors (Amph), but not eMMLV receptors (Eco) or the CAR. Therefore, MO7-e cells lack tropism for eMMLV and adenovirus. To form targeted hybrid conjugate vectors, individual components were either biotinylated or avidinylated and assembled in a stepwise fashion.

FIG. 2.

Formation of retrovirus-adenovirus conjugates enhances retroviral transduction efficiency in MO7-e cells and enables c-kit-targeted retrovirus transduction. MO7-e cells were incubated with biotinylated EGFP encoding amphotropic retrovirus at a constant multiplicity of infection (MOI) of 2 for each data point. c-kit receptor retargeting was accomplished via linkage of biotinylated retrovirus to SCFbiot via a neutravidin bridge. Untargeted constructs were assembled with plain (unbiotinylated) SCF (no bridge formation). Addition of avidinylated adenovirus established retrovirus-adenovirus conjugates, which were tested with or without c-kit targeting. Avidinylated adenovirus was incorporated at a different MOI into the conjugate. Open bars indicate use of unmodified (incapable of bridge formation) adenovirus at an MOI of 10 (A). The experiments were also conducted with biotinylated ecotropic retrovirus (B). ND, not detected. Green fluorescent cells were measured by fluorescence-activated cell sorter and plotted as percentages of total cells. Data points represent means ± standard errors of the means from quadruplicate experiments.

Formation of ecotropic MMLV-adenovirus conjugates enables c-kit-targeted retroviral transduction.

To demonstrate specific vector uptake exclusively via the c-kit receptor, different ecotropic retrovirus conjugate vectors encoding the EGFP reporter gene were synthesized using the biotin-streptavidin technique. AdCMVLuc-Av was incorporated for endosomalysis, and the SCF moiety was incorporated for targeting, as depicted in Fig. 1. The ecotropic retrovirus by itself did not transduce human MO7-e cells, and conjugation of eMMLV with an adenovirus also failed to result in transduction. Considering that MO7-e cells lack the CAR (40), this is not an unexpected result (Fig. 2B). The retargeting of the ecotropic retrovirus by linkage to SCF also failed to result in transduction, a finding consistent with previous reports by Yajima et al. (49) and Fielding et al. (15). However, the physical conjugation of an adenovirus to the c-kit-targeted retrovirus provided dose-dependent retroviral cellular transduction. The transduction efficiency was similar to that obtained with the vector constructs assembled with the amphotropic retrovirus (compare Fig. 2B with Fig. 2A). These experiments demonstrate that a c-kit-redirected retrovirus in conjunction with an endosomalytic adenovirus can effectively transduce host cells. They suggest that particles internalized via the c-kit receptor are preferentially processed through the endosomal/lysosomal pathway. Consistent with this conclusion is a previous report from our group that the introduction of an adenovirus into a c-kit-targeted molecular conjugate vector enhanced the gene expression of a reporter plasmid 2,000-fold (43).

CD34-positive human PBSC express the CAR only at minimal levels.

Reports on gene transfer to primary HSC by means of an unmodified adenovirus have been variable, from no significant gene transfer (11) to gene expression in up to 20% of HSC preparations (16). These conflicting findings could be explained by the heterogeneity of human HSC collections. Given its inherent endosomalytic properties, a retargeted recombinant adenovirus could be a model vector with which to study the feasibility of c-kit-mediated gene transfer in HSC. To exclude possible direct adenovirus-mediated uptake, CD34⁺-selected HSC were examined for CAR expression by flow cytometry (4). In PBSC, only 4.5% of the total population were found to be positive for CAR, with very low CAR expression in CD34⁺ gated cells (<0.2%) (Table 1). We established an inverse correlation of CAR expression with the CD34 epitope in HSC, with CAR being extremely low on CD34^bright cells. Back-gating confirmed that CAR expression was limited mainly to larger cells outside the “stem cell gate.” These results suggest that CD34^bright HSC are poor targets for CAR-dependent, untargeted adenovirus vector-mediated gene transfer. In contrast, A549 cells, a cell line that can be readily transduced with an adenovirus, were 100% positive for CAR expression (Table 1).

TABLE 1.

Human CD34⁺ PBSC do not express CAR and are poorly infected by recombinant adenovirus^a

Cell type	% of cells positive for:		CAR labeling-induced MCFb shift
	EGFP	CAR
CD34 gated	<0.1	<0.2	7 to 7
CD34 total population	6	4.5	8 to 21
A549	100	100	11 to 275

^aA549 and CD34-selected PBSC were infected with Ad-EGFP and analyzed for EGFP expression by flow cytometry as outlined in Materials and Methods. Cells were also analyzed for CD34 and CAR expression. 

^bMCF, mean channel fluorescence. 

c-kit-retargeted recombinant adenovirus results in high-efficiency gene transfer and gene expression in CD34-positive human PBSC.

Cells were transduced with control vector (SCFbiot + Ad-EGFP) and the c-kit-targeted construct consisting of avidinylated adenovirus (SCFbiot + Ad-EGFP-Av). With the targeted vector, 66.3 ± 0.4% of all cells and 28.3 ± 3.2% of CD34^bright cells expressed EGFP, whereas only 5.5 ± 1.3% of all cells and 2.5 ± 0.7% of CD34^bright cells expressed EGFP with the control vector (P < 0.0001) (Fig. 3A). When the targeted vector was used, mean channel fluorescence increased over that in control vector-transfected cells from 205 ± 30 to 315 ± 6 (total cell population) and from 723 ± 30 to 976 ± 45 (gated cell population) (P < 0.001) (Fig. 3B). No gene expression was seen in CD34^bright cells (CAR^low) after incubation with Ad-EGFP or the control vector Ad-EGFP-Av. The results were consistent in two consecutive and independent experiments. These data correlate with previous findings reported by Chen et al., who did not observe adenovirus vector-mediated transgene expression in human CD34⁺ HSC (11). In contrast, the detection of gene expression in CD34⁺ HSC reported by Frey et al. (16) could be explained by donor differences similar to differences in transduction efficiency reported with adeno-associated virus in bone marrow cells (36). Although a shift in mean channel fluorescence was seen in the total HSC population with the control vector Ad-EGFP, this was limited to cells outside the CD34 gate. However, a shift in mean channel fluorescence was also seen in both the total cell population and the CD34-gated population with the c-kit-directed construct. This may be explained by the fact that c-kit can also be present on less-primitive CD34^low hematopoietic precursor cells (29). In summary, c-kit retargeting of recombinant adenovirus enables effective gene transfer to primitive human CD34⁺ HSC.

FIG. 3.

c-kit-targeted recombinant adenovirus efficiently transduces CD34-selected human PBSC. Primary, CD34-selected human stem cells were transfected with a control vector (SCFbiot + Ad-EGFP) or a retargeted vector (SCFbiot + Ad-EGFP-Av) as outlined in Materials and Methods. (A) Cells were analyzed by flow cytometry for EGFP expression. Solid bars, percentage of total cells expressing EGFP; open bars, percentage of CD34^bright cells expressing EGFP. Data points represent results from triplicate experiments. (B) Mean channel fluorescence of the total cell population(solid bars) and of cells contained within the CD34^bright gate (open bars).

CD34⁺-selected human stem cells can effectively internalize targeted vector constructs via the c-kit receptor.

To exclude a CAR-independent pathway of adenovirus transduction in primary human CD34⁺ HSC, studies were performed with Cy3-labeled adenovirus. Cells were then analyzed by confocal microscopy for vector binding (4°C) and internalization (37°C). CD34⁺-selected HSC were incubated with Cy3-labeled adenovirus (SCF + Ad-Cy3-Av) (Fig. 4B1 and B2) or c-kit-retargeted vector (SCFbiot + Ad-Cy3-Av) (Fig. 4C1 and C2). For comparison, A549 cells were infected with Cy3-labeled adenovirus (Fig. 4A1 and A2). On A549 cells incubated with Ad-Cy3-Av, high-density cell membrane attachment was seen at 4°C, with efficient and complete internalization at 37°C (Fig. 4A1 and A2, respectively). As predicted from the CAR expression studies, uptake of untargeted adenovirus in A549 cells was very efficient, compared to only scarce uptake in primary HSC. However, adenovirus c-kit targeting resulted in highly efficient vector labeling and internalization in primary human HSC, indicating that adenovirus restricted tropism could be overcome via c-kit retargeting. PBSC were poorly labeled with Ad-Cy3, and internalization was observed in less than 1% of cells (Fig. 4B1 and B2, respectively). With the c-kit-retargeted vector (SCFbiot + Ad-Cy3-Av), both effective cell-surface labeling and significantly enhanced cytoplasmic uptake were observed in PBSC (Fig. 4C1 and C2, respectively).

FIG. 4.

CD34-selected human stem cells effectively internalize the c-kit-targeted recombinant adenovirus gene transfer vector. A549 cells and PBSC were infected with Cy3-labeled adenovirus at a multiplicity of infection of 100 (A and B, respectively), and PBSC were transfected with the c-kit-targeted, Cy3-labeled construct SCFbiot + Ad-Av-Cy3 (C). Cells were analyzed by confocal microscopy. Cells were maintained for 1 h at 4°C for demonstration of cell surface binding (A1, B1, and C1) or at 37°C for demonstration of vector internalization (A2, B2, and C2).

c-kit-retargeted recombinant adenovirus enters cells specifically via the c-kit receptor.

Previously, our group reported c-kit targeting in cell line studies using a c-kit-targeted adenovirus-polylysine conjugate (43). This vector did not confer specific c-kit targeting in primary CD34⁺ cells, because in this cell population uptake occurred preferentially via the polylysine component and not via c-kit (reference 40; also unpublished data). To demonstrate the specificity of the redirected adenovirus vector and its exclusive uptake via c-kit, HSC were pretreated with the c-kit blocking antibody SR1 (34, 43). CD34-selected human PBSC were transfected with the c-kit-targeted adenovirus vector (SCFbiot + Ad-EGFP-Av). Cells were preincubated with the c-kit-blocking monoclonal antibody SR1 or its isotype control at 2 μg/ml (34). Eighty-one percent of cells expressed EGFP with the targeted vector, 30.4% expressed EGFP with SR1 pretreatment, and 75% expressed EGFPwith isotype control treatment (Fig. 5). These results were confirmed twice.

FIG. 5.

Blocking of c-kit reduces transfection efficiency with the c-kit-targeted vector. CD34-selected primary human stem cells either were not pretreated or were preincubated with the SR1 antibody or its isotype control at 2 μg/ml prior to transfection with the c-kit targeting vector (SCFbiot. + Ad-EGFP-Av). Results are expressed as percent cells expressing the EGFP transgene.

Taken together, our results clearly indicate the feasibility of c-kit-targeted gene transfer in this ultimate target cell population. Transient gene expression with an adenovirus in HSC could become a very useful procedure, for instance, for temporarily conferring drug resistance to accomplish chemotherapy or radiation protection (42, 45). Nevertheless, other disorders do require stable transgene expression of HSC for cure (14, 21, 23, 45, 47). Our data and the findings of Yajima et al. (49) suggest that the design of novel retrovirus-integrating, c-kit-targeted vectors requires the incorporation of endosomalytic agents. Strategies for designing such vectors are under way with the development of adenovirus-retrovirusl hybrid vectors (6, 37). Alternatively, endosomalytic peptides could be engineered to be expressed on c-kit-targeted retroviruses (35).