Sustained Gene Expression in Retrovirally Transduced, Engrafting Human Hematopoietic Stem Cells and Their Lympho-Myeloid Progeny

The LXSN type of MoMLV vector was used as the parental vector for LINGFR.23 After deleting the internal SV40 promoter and theneo gene in LXSN, an internal ribosome entry site (IRES) from the encephalomyocarditis virus was inserted.24 The human (p75) NGFR gene was then placed after the IRES.25 The resultant vector was named LINGFR to reflect the order of essential components (LTR-IRES-NGFR). The MINGFR vector was similarly constructed by replacing the neo gene driven by an internal promoter in MSCVneoEB with the IRES-NGFR cassette from LINGFR.20 Gene expression mediated by the MINGFR vector is directed by the MSCV LTR, whereas in LINGFR gene expression is driven by the MoMLV LTR. A third vector, MINT, was created by truncating the NGFR gene in MINGFR after the transmembrane domain (at the Nae I site). All the plasmids were purified and used in packaging cell transfections as described.26 

The primers used to amplify transgene-specific (IRES) DNA sequences common to all three vectors were: upstream, 5′-CGT TAC TGG CCG AAG CCG CT-3′; and downstream, 5′-AAC CTC GAC TAA ACA CAT GT-3′. The primers used to amplify the endogenous human β-globin sequence of genomic DNA as a control for PCR assays were: upstream, 5′-ACA CAA CTG TGT TCA CTA GC-3′; and downstream, 5′-CAA CTT CAT CCA CGT TCA CC-3′. Forty-cycle PCR reactions for both target sequences were performed with an annealing temperature of 62°C in the presence of 1.5 mmol/L MgCl₂. PCR products (a 485-bp IRES-specific fragment and a 220-bp β-globin–specific fragment) were separated by 4% agarose gel electrophoresis.

A hybridoma producing mouse IgG1 monoclonal antibody (MoAb) against the human NGFR was obtained from the American Type Culture Collection (ATCC HB8737; Rockville, MD). Purified antibodies from mouse ascites were conjugated either directly with fluorescein isothiocyanate (FITC) or with R-phycoerythrin (PE) after deleting the Fc fragment. A CD34 antibody (Tuk3) conjugated with sulfo-rhodamine (SR), and a panel of FITC-conjugated mouse MoAbs against lineage differentiation markers were used to isolate and analyze transduced cells.27 This lineage panel (collectively called Lin) comprised CD2, CD4, CD14, CD15, CD16, CD19 (Becton Dickinson, San Jose, CA) and glycophorin A (Immunotech, Westbrook, ME). An FITC-conjugated antibody (MA2.1, ATCC HB54) against HLA-A2 was used to identify and monitor MA2.1⁺ donor cells in SCID-hu bone mice.12 

Propidium iodide (0.5 μg/mL) was added to cell suspensions after antibody staining to exclude dead/dying cells from FACS analyses. FACStar^Plus or FACS Vantage cell sorters (Becton Dickinson) equipped with a primary Agron laser and a dye-laser (required for detecting SR signals) were used for cell sorting. FACScan analyzers (Becton Dickinson) equipped only with an Argon laser were used to phenotype harvested cells after in vitro and in vivo assays.

Recombinant human interleukin-3 (IL-3), IL-6, granulocyte-macrophage colony-stimulating factor (GM-CSF), steel factor (SLF, also called stem cell factor), and leukemia inhibitory factor (LIF) were obtained from Sandoz Pharma (Basel, Switzerland), and erythropoietin (Epo) was purchased from Amgen (Thousand Oaks, CA). Dulbecco's modified Eagle's medium (DMEM), Iscove's modified Dulbecco's medium (IMDM), and RPMI 1640 culture media were purchased from GIBCO-BRL (Gaithersburg, MD) and fetal calf serum (FCS) from Hyclone (Logan, UT). TF1 cells (ATCC CRL-2003) were maintained in RPMI 1640 plus 10% FCS and 2 ng/mL GM-CSF.28 

Amphotropic supernatants produced by the human 293 (embryonic kidney fibroblast) cell-based ProPak-A packaging line were made through transduction with VSV-G pseudotyped viral stocks as described.29 30 NGFR-expressing ProPak-A cells were enriched by flow cytometry sorting and expanded in culture. Amphotropic supernatants were then collected from stable ProPak-A producers, filtered, and stored at −80°C until use. For transduction, fresh or previously frozen vector supernatants were mixed at a 1:1 ratio with media containing target cells in the presence of 8 μg/mL polybrene (Sigma, St Louis, MO). The transduction mixture was then centrifuged at 1,800g at 32°C to 35°C for inoculation. After the 4-hour “spinoculation,” the cells were washed once and cultured in appropriate media.

Human BM aspirates and mobilized peripheral blood (mPB, collected at day 4 or 5 after G-CSF treatment) were obtained from healthy donors in compliance with regulations established by the federal and state governments. Low-density (<1.077 g/cm³) mononuclear BM cells after Ficoll-Hypaque gradient (Pharmacia, Piscataway, NJ) or apheresed mPB cells were stained with a CD34 antibody included in the Isolex kit (Baxter Biotech Immunotherapy Division, Irving, CA). CD34⁺ cells were magnetically isolated by Isolex using a modified protocol developed at Systemix.27 The purity of CD34⁺ cells from both BM and mPB was usually greater than 90% (n = 10 for BM and n = 20 for mPB). These isolated cells were then stained with CD34-SR and Lin-FITC, and CD34⁺Lin⁻ cells were sorted by FACS and activated ex vivo for gene transduction. Cells (≈10⁶/mL) were cultured overnight in IMDM/RPMI 1640 (1:1) medium plus 10% FCS supplemented with 10 ng/mL IL-3 and IL-6, and 100 ng/mL SLF. The next day cells were transduced with ProPak-A viral supernatants for 4 hours as described above. The transduction procedure was repeated the following day, and then the medium was changed and the cells were cultured for additional 2 days to allow gene expression.

Methylcellulose and reagents for clonogenic progenitor assays were obtained from StemCell Technologies (Vancouver, Canada). Cells (4 × 10²) were added to 1 mL methylcellulose medium (MethoCult H4230) supplemented with IL-3, IL-6, GM-CSF (10 ng/mL each), SLF (100 ng/mL), and Epo (2 U/mL). Cell mixtures were plated in 35-mm suspension culture dishes (Nunc, Roskilde, Denmark), and incubated at 37°C. After 2 weeks, colonies (>100 cells) were enumerated in each of three triplicate plates. Subsequently some individual colonies were randomly picked. Cells were then lysed and an aliquot was used for a 40-cycle PCR analysis to detect specific DNA sequences. Colonies which yielded a positive signal for the human β-globin sequence were included in the calculations of colony-forming cell (CFC) gene transfer efficiency. Colonies which also yielded a transgene signal equal to or stronger than the β-globin signal in the same PCR reaction were considered to be positive for the transgene. The remaining cells were obtained in bulk at the end of 2-week CFC assays for FACS analysis of NGFR expression. Cells were washed three times with phosphate-buffered saline plus 0.5% human IgG (Gammimmune, Miles Inc, Elkhart, IN) before being stained with a PE-conjugated NGFR antibody and FITC-conjugated antibodies against glycophorin A, CD14, or CD15.

For stromal-dependent cobblestone area-forming cell (CAFC) assays,31 a stromal cell line (SyS-1) derived from murine BM was used.13 Cultures were maintained with IMDM/RPMI 1640 (1:1) medium plus 10% FCS supplemented with 10 ng/mL IL-6 and 50 ng/mL LIF. Up to 100 cells were cultured on sub-confluent monolayers of SyS-1 stromal cells in each well of 96-well plates. Half of the medium was replaced weekly, and the cultures were monitored for 6 weeks. CAFC frequencies were calculated based on the results obtained at week 5. Total cells were procured and pooled from wells which contained at least one cobblestone area at week 6. After being filtered through a 30-μm mesh filter, cells were stained as described before, with the PE-conjugated NGFR antibody and the FITC-conjugated CD19 antibody.

Immunodeficient C.B-17 scid/scid (SCID) mice were used as recipients of human fetal bone fragments to construct the SCID-hu bone mice.11,12 In each of two independent experiments, bone fragments were derived from the same human fetal tissue, which was negative for HLA-A2 and B17 and was not recognized by the corresponding MoAb MA2.1 (ATCC HB-54). Eight weeks after bone implantation, SCID-hu bone mice were used as recipients of transduced human hematopoietic precursors whose HLA-type was MA2.1⁺. The transduced donor cells were injected directly into the implanted bones (MA2.1⁻) residing in SCID mice according to the published protocol.12 Fifty thousand to 100,000 cells (based on CD34⁺Lin⁻ cell counts) from the same cell population were injected into each of two bone fragments implanted in SCID mice. Nine weeks after injection, recipient mice were terminated, the implanted bone fragments were surgically removed, and total cells from BM were obtained. Then harvested cells were then stained with the PE-conjugated NGFR antibody and the FITC-conjugated MA2.1 antibody. Propidium iodide was added after antibody staining to cell suspensions to exclude dead/dying cells from FACS analyses. Using a FACScan analyzer, levels of engraftment (based on the presence of MA2.1⁺ donor cells) and transgene expression (based on the presence of NGFR on the surface of donor cells) were assessed by FACS.

Several retroviral vectors containing the NGFR gene as a reporter were constructed to compare gene transfer and expression in human hematopoietic cells (Fig 1). NGFR expression is directed by the MSCV LTR in MINGFR and MINT, or directed by the MoMLV LTR in the LXSN-derived vector LINGFR.20,23 To achieve a higher efficiency of transduction of human hematopoietic precursors than has been routinely accomplished with recombinant retroviral stocks prepared using conventional murine packaging lines, amphotropic retroviral vector supernatants were produced using a new human 293 cell-based packaging line (ProPak-A).29Transduction efficiencies were first assessed on the growth factor–dependent human CD34⁺ progenitor line TF1.28 As shown in Fig 2A, ≈60% of TF1 cells were transduced by either MINGFR or LINGFR vector 4 days after the cells were exposed to 50% (vol/vol) of viral supernatants in a “spinoculation” protocol. Subsequent limiting dilution experiments with MINGFR and LINGFR vector preparations established that the two viral stocks had comparable titers (data not shown). Moreover, as indicated by the mean fluorescence intensities of the positive peaks, both vectors directed similar NGFR expression levels in TF1 cells (Fig 2A). The two vectors also performed equally well when tested on other cell lines such as murine NIH3T3 fibroblasts (data not shown).

A third vector, MINT, in which the NGFR gene in the MINGFR vector was C-terminally truncated after the transmembrane domain, was used to examine the kinetics of NGFR expression after viral transduction. The titer of MINT viral stock was somewhat lower and ≈40% TF1 cells were transduced and expressed NGFR on cell surface after 4 days (Fig 2A). To determine the kinetics of NGFR transgene expression after transduction, MINT-transduced TF1 cells were either stained for NGFR surface protein immediately posttransduction or cultured for various periods of time and then analyzed as described above. No NGFR signal could be detected on the cell surface immediately after the transduction protocol (Fig2B). Subsequently, the percentage of NGFR-expressing (NGFR⁺) cells as well as NGFR signal intensities progressively increased with time, reaching a maximum of 40% 4 days posttransduction (Fig 2A and B). Afterward the values of these two parameters remained constant for at least 2 months if transduced TF1 cells were maintained in a proliferating phase. The doubling time of TF1 cells is ≈20 hours; thus, it took a period equivalent to three cell divisions for continuously proliferating TF1 cells to reach the maximal and stable level of the NGFR surface signal. This finding is consistent with accumulated data that efficient retroviral-mediated expression (transcription and translation) occurs only after integration into cellular chromosomes which, in the case of C-type retroviruses, is dependent on cell division.

The MINT vector was initially constructed to minimize the possibility of NGFR functioning as a signal transducer in hematopoietic cells. However, no adverse effect was observed due to full-length NGFR expression on the growth of transduced TF1 cells (data not shown) or primary human hematopoietic cells assayed in vitro (see below). Because the titer of the MINT vector was somewhat lower than that of MINGFR or LINGFR vector (Fig 2A), we restricted our attention solely to the latter two vectors to compare directly the expression properties of MSCV- and MoMLV-based retroviral vector backbones after transduction of primary human hematopoietic precursors.

Enriched CD34⁺ cells from BM or mPB were further purified by FACS to reach greater than 95% purity with respect to CD34 expression and lack of expression of lineage-specific differentiation markers (CD2 and CD4 for T cells, CD16 for natural killer [NK] cells, CD19 for B cells, CD14 for monocytes, CD15 for granulocytes, and glycophorin A for erythroid cells). These highly purified cell populations (denoted as CD34⁺Lin⁻) were prestimulated in a cytokine cocktail (IL-3, IL-6 plus SLF) overnight and then exposed to retroviral supernatants in the “spinoculation” protocol. Viral supernatants were added on days 1 and 2, and then the cells were cultured for 2 more days to allow gene expression. Gene transfer efficiency and NGFR expression level were then investigated by FACS analysis. Under the conditions used, total cell numbers increased during this period by approximately fourfold or twofold for input CD34⁺Lin⁻ cells isolated from BM (n = 3) or mPB (n = 6), respectively. Shown in Fig 3 as an example, ≈80% of the cells retained the CD34⁺Lin⁻ phenotype after 4 days in culture while the remaining cells lost the CD34 antigen and gained one or more lineage differentiation markers (n = 3 for mPB). The CD34⁺ content of the cultures decreased rapidly thereafter, concomitant with the continued increase in total cell numbers. Among mPB cells which retained the CD34⁺Lin⁻phenotype (gated in the R2 region in Fig 3), greater than 30% expressed the NGFR transgene at day 4, irrespective of which retroviral vector backbone was used. The percentages of NGFR⁺ cells and the corresponding mean fluorescence intensities were, however, marginally but consistently higher for cells transduced by MINGFR than for those transduced by LINGFR (n = 3 for mPB). Similar results were obtained with transduced CD34⁺Lin-cells of BM origin (n = 2, data not shown).

At day 2 posttransduction (4 days in culture), cells were plated in methylcellulose to detect clonogenic myeloid and erythroid progenitors in a standard CFC assay. Total numbers of colonies were first enumerated by visual inspection and then the efficiency of gene transduction was assessed either by PCR to detect the presence of the vector sequences in individually picked colonies or by FACS to examine NGFR expression in populations of CFC-derived myeloid and erythroid cells. Approximately 25% of the input cells transduced by either MINGFR or LINGFR formed colonies, similar to nontransduced controls (Table 1). PCR analysis indicated that 28% (out of 64 picked CFC colonies) were transduced by MINGFR and 34% (out of 65 picked CFC colonies) were transduced by LINGFR, indicating that both vectors were capable of achieving similar efficiencies of gene transfer to CFC. When NGFR expression in differentiated cells derived from CFC was analyzed after 2 weeks in culture, we found that only 7.7% to 12.5% of the total cell populations expressed detectable cell surface NGFR (Table 1).

We next sorted transduced CD34⁺Lin⁻ cells into two fractions on the basis of NGFR expression at day 2 posttransduction (Fig 3). The fact that a small percentage of NGFR⁻ cells were capable of generating NGFR⁺ differentiated cells in the CFC assays (by FACS, data not shown) illustrated that not all of the retrovially transduced cells expressed sufficient levels of NGFR transgene at day 2 posttransduction to permit their identification and isolation. Thus, the frequencies of NGFR⁺ cells at this time point (day 2 posttransduction) were presumed to be underestimates of the actual values of cells containing the transgene. Nonetheless, since enrichment for cell-surface NGFR greatly simplified functional analyses of transduced cells and studies of LTR-mediated gene expression in functionally heterogeneous CD34⁺Lin⁻ cells, we subsequently focused on those NGFR⁺ subpopulations that were clearly transduced and could be separated from untransduced cells. Approximately 20% of the CD34⁺Lin⁻NGFR⁺ cells generated by transduction with MINGFR or LINGFR formed CFC colonies (Table 1), a rate slightly less than that of unsorted cells (which is ≈25%). Based on these plating efficiencies (20% v 25%), this result would indicate that the majority of transduced CFC progenitors by both vectors also expressed the NGFR transgene 2 days posttransduction. Interestingly, however, NGFR transgene expression in CFC-derived cells at the end of the 2 week CFC assay period are different among NGFR⁺ cells transduced by the two vectors. Whereas 78% of CFC-derived progeny cells derived from MINGFR-transduced, sorted CD34⁺Lin⁻NGFR⁺ cells continued to express the NGFR transgene after 2 weeks, approximately one half of cells derived from LINGFR-transduced NGFR⁺ CFC had completely lost NGFR transgene expression while the remaining expressed NGFR at a reduced level (Table 1 and Fig 4). PCR analysis after cell sorting of LINGFR-transduced, CFC-derived cells which did not express NGFR after 2-week CFC assays confirmed that the transgene was still present in this NGFR⁻ differentiated cell population (data not shown). Further multiparameter FACS analyses shown in Fig 4 indicated that downregulation of LINGFR-mediated NGFR expression occurred in differentiated erythroid (glycophorin A⁺) cells and granulocytes (CD15⁺), as well as in (CD14⁺cells) monocytes (data not shown). The downregulation of MoMLV LTR-mediated NGFR expression by the LINGFR vector in CD14⁺and CD15⁺ myeloid cells was also seen in a suspension culture assay (data not shown). These findings are consistent with those of a recent report in which the MoMLV LTR-mediated transcription is downregulated in differentiated erythroid/myeloid progeny derived from transduced CD34⁺ cells in CFC and suspension cultures, and the MSCV LTR-mediated transcription is substantially higher.19 

Because silencing of MoMLV LTR-mediated expression had previously been observed in resting human T cells,17 and in differentiated erythroid/myeloid cells as shown above, we next examined whether this phenomenon also occurred in differentiated CD19⁺ B-lineage cells. The sorted NGFR⁺ and NGFR⁻fractions of CD34⁺Lin⁻ cells (lacking CD19 expression) were plated at limiting dilution on stromal cell monolayers in a CAFC assay. There are two unique aspects of this long-term stromal-dependent CAFC assay compared with CFC assays: (1) CD19⁺ B-lineage cells as well as myeloid cells are generated from CD34⁺Lin⁻ cells; and (2) late-appearing (after 5 weeks) cobblestone areas are considered to be derived from progenitors which are more primitive than CFC.13 31 The frequencies of CAFC in the various transduced cell populations were calculated based on Poisson distribution (Fig 5A). The frequencies of CAFC present in the NGFR⁺ subpopulations transduced by the MINGFR and LINGFR vectors were similar. However, both frequencies were more than 10-fold lower than the values calculated for the corresponding NGFR⁻ subpopulations, indicating that the majority of CAFC were either not transduced at all or not expressing NGFR transgene 2 days posttransduction. An additional experiment using different preparations of viral supernatants confirmed this finding (data not shown).

NGFR transgene expression was examined in differentiated B cells at week 6 of CAFC assays. Cells in cobblestone area–containing wells initially seeded with preselected NGFR⁺ transduced cells were obtained and pooled for simultaneous FACS analysis of the CD19 B-cell marker and NGFR transgene expression (Fig 5B). The majority of harvested hematopoietic cells were myeloid cells. A population of CD19⁺ cells was detected in pooled CAFC⁺ wells containing transduced cells by either LINGFR or MINGFR vector, or mock-transduced cells (Fig 5B). Although ≈50% of CD19⁺cells derived from MINGFR-transduced CD34⁺Lin⁻NGFR⁺ cells expressed a high level of NGFR reporter, a low to medium level of cell-surface NGFR could only be detected on ≈14% of CD19⁺ cells derived from LINGFR-transduced CD34⁺Lin⁻NGFR⁺ cells. Taken together, the results show that the MSCV LTR appears to be less susceptible to transcriptional silencing mechanisms than the MoMLV LTR in multiple lineages during in vitro differentiation of transduced human hematopoietic precursors.

It has been shown that subsets of CD34⁺Lin⁻ cells capable of long-term (>8 weeks) engraftment of fetal human bone implants in SCID mice (SCID-hu bone assay) exhibit B-lymphoid and myeloid potential as well as secondary repopulating capacity.12 13 To assess whether transduced CD34⁺Lin⁻ cells expressing the NGFR reporter have candidate HSC activity, transduced mPB cells were tested in the SCID-hu bone assay for the presence of SCID repopulating cells (SRC). Because NGFR expression is higher in progeny derived from CD34⁺Lin⁻ cells transduced with MINGFR than with LINGFR in the in vitro assays described above, for SRC assays we restricted our attention to the MINGFR-transduced cells. Nine weeks postinjection, the presence of donor (MA2.1⁺) cells and the NGFR transgene expression was examined by FACS analysis of total cells obtained from the human bone implants (Fig6). Cell-surface expression of NGFR could not be detected in nontransduced human hematopoietic cells (Fig 6A and B). The sorted CD34⁺Lin⁻NGFR⁺ cells that had been transduced by the MINGFR vector readily engrafted and maintained NGFR transgene expression in donor cells 9 weeks postinjection (Fig 6C and D).

The results of our first two experiments are summarized in Table 2. Nine of 10 injections with mock-transduced cells showed donor cell engraftment. Because this frequency (90%) is similar to historic data obtained using freshly isolated human hematopoietic precursors,12 we believe that our protocol for ex vivo cell culture and transduction did not significantly alter the long-term SRC potential of CD34⁺Lin⁻ cells. Both NGFR⁺and NGFR⁻ subpopulations selected at day 2 posttransduction by MINGFR engrafted (5 of 5 and 2 of 2, respectively). All 5 bone grafts that successfully repopulated with MINGFR-transduced, preselected CD34⁺Lin⁻NGFR⁺cells showed reasonable levels of NGFR transgene expression (28% to 45% [37% ± 6%, n = 5] of the donor cells). In one experiment where sufficient numbers of donor-derived cells were available for an additional FACS analysis, both CD19⁺ (B-lineage) and CD19⁻NGFR⁺ (transduced donor) cells were found (data not shown). These findings showed that the MSCV LTR is functional in primitive human hematopoietic precursors with SRC potential and remains active for long periods in their transduced progeny after in vivo differentiation.

In this report we evaluated the MSCV-based MINGFR vector carrying the NGFR reporter gene for its ability to efficiently transduce hematopoietic precursors purified from adult human BM or mPB. Based on FACS analyses and in vitro functional assays, similar efficiencies of gene transfer into CD34⁺Lin⁻ cells as well as CAFC and CFC progenitors were observed as for the MoMLV-based LINGFR vector. Moreover, stable gene transfer into HSC (or SRC), based on the NGFR reporter expression in engrafted donor cells, was observed in 7 of 7 grafts of MINGFR-transduced cells. Because of the qualitative nature of the SCID-hu mouse model (in the absence of a limiting dilution analysis which requires large numbers of engineered animals), it is difficult to estimate the frequencies of HSC/SRC-based gene transfer by MINGFR, and to what extent the improved retroviral vector stocks and the “spinoculation” transduction protocol used in this study contributed to the success of these experiments. Because others reported recently that inclusion of certain cytokines (eg, Flk2/Flt3 ligand) and the presence of cell extra-cellular matrix molecules (eg, fibronectin fragments) can further preserve/activate candidate HSC and increase efficiencies of gene transfer into them, we expect that the transduction protocol reported here can be further optimized with these molecules.14,32 33 

Although the level of NGFR expression directed by the LTR of either vector was comparable in several cell lines and in the total CD34⁺Lin⁻ cell populations shortly after transduction, NGFR transgene expression mediated by MSCV LTR is substantially higher than that directed by MoMLV LTR in differentiated progeny derived from transduced CD34⁺Lin⁻cells (Figs 4 and 5B). We observed that the LTR-mediated gene expression from LINGFR (which is based on the LXSN-type of MoMLV vector) was downregulated in differentiated cells belonging to multiple-lineages including erythroid, myeloid, and B-lymphoid cells. Similar observations with MoMLV vectors have been made by other investigators in erythroid/myeloid lineages,19 in T cells,17 and in mouse hematopoietic cells after in vivo BM repopulation.18 Moreover, the phenomenon of the in vivo downregulation of gene expression of MoMLV LTR (typically caused by transcriptional silencing/inactivation) is not limited to the hematopoietic system, as it has also been observed in transduced primary fibroblasts, myoblasts, and hepatocytes in a variety of animal models.34 However, it should be noted that others have documented MoMLV LTR-directed gene expression in the T-cell or myeloid progeny of transduced CD34⁺ cells purified from cord blood or fetal liver after repopulation of SCID-hu thymus or SCID-hu bone grafts.35-37 Nonetheless, the possible extinction of MoMLV LTR-mediated expression needs to be taken into account when sustained expression in multiple myeloid and lymphoid lineages is deemed necessary for therapeutic benefits in HSC-based gene transfer applications.

Based on the outcome of this study, it would appear that MSCV-based vectors may offer advantages over conventional MoMLV-based vectors for gene delivery to the human hematopoietic system. The ability to direct sustained high-level expression of exogenous genes in differentiated cells derived from transduced HSC/progenitors is obviously a desirable goal of a number of human gene therapy protocols targeting congenital blood disorders.1 In addition, a vector that is permissive for expression in HSC may be of utility in those cancer gene therapy applications where the intent is to confer a drug-resistant phenotype to the patient's mature hematopoietic cells and their precusors to augment the therapeutic index of high-dose anti-tumor chemotherapy. In this regard, other types of retroviral vectors are currently being investigated for this purpose and are promising in human progenitor cells assayed in vitro.38 It remains to be determined whether these vectors are also functional in directing gene expression in HSC/SRC and whether they are more active than MSCV.

The sensitivity of the NGFR reporter system allowed facile monitoring of transgene expression during differentiation of transduced human hematopoietic precursors into progeny cells belonging to multiple (myeloid and lymphoid) lineages.22 Because the NGFR reporter gene is of human origin, it provides advantages in terms of increased specificity in SCID mouse models (compared with other mouse reporter genes in use) and, presumably, reduced immunogenicity in humans (compared with bacterial or mouse reporter genes). It was a concern at the outset that endogenous NGFR expression in human hematopoietic cells may complicate the SCID-hu bone assay. However, we have not been able to detect cell-surface NGFR expression in primary CD34⁺ cells, their progeny cells differentiated in vitro and in vivo, or in fetal hematopoietic cells residing in the recipient bone fragments in any of our experiments. In any case, although the presence of the full-length NGFR gene did not exert any noticeable adverse effects on the transduced hematopoietic precursors we evaluated, use of vectors like MINT expressing C-terminally truncated NGFR genes should alleviate any remaining misgivings. Therefore, the stable and nonimmunogenic NGFR reporter coexpressed with a therapeutic gene from the same vector may be useful in human gene therapy based on human HSC as well as T cells,39 if an easily detectable marker is desired to monitor and isolate transduced cells.

In summary, MSCV-based retroviral vectors encoding easily detectable and selectable markers should facilitate studies aimed at further characterizing the CD34⁺Lin⁻ subset containing SRC. It is anticipated that the information gained will lead to improvements in HSC-based gene and cellular therapies.

We are grateful to Dr Richard Rigg for providing ProGag and ProPak-A packaging cells, and to Dr Ivan Plavec for providing a LXSN-based retroviral vector and an NGFR-containing plasmid. We also thank the SyStemix Cell Processing Group for isolation of CD34⁺ cells and the Comparative Medicine Group for production of SCID-hu mice.