INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

Gene therapy of many of the most important therapeutic targets will require transduction of nondividing cell types, including hematopoietic progenitors. Most retroviruses, including murine leukemia virus (MuLV), cannot complete a replicative cycle in nondividing cells because the preintegration complex is unable to traverse the intact nuclear membrane (23, 35). However, human immunodeficiency virus type 1 (HIV-1) has at least two gene products which allow nuclear entry in resting cells. The first is matrix (MA), which is located at the N terminus of Gag and has a canonical nuclear localization signal; in the absence of Vpr, it is required for efficient replication of HIV in primary human macrophages (2, 10, 11, 43). MA has been shown to interact biochemically with alpha-importin (karyopherin-1), which may be partly responsible for docking the preintegration complex at the nuclear pore (9). The second gene product is Vpr, which has alpha-helices which are required for nuclear localization, and in the absence of MA it is sufficient for HIV replication in primary cells (16). Because of this property of HIV-1 (and presumably of the related lentiviruses such as simian immunodeficiency virus [SIV] and HIV-2), HIV-1 vectors have been used to transduce human cells. Akkina et al. used a replication-defective HIV pseudotyped with vesicular stomatitis virus G (VSV G) protein to transduce CD34⁺ cells (1). However, they did not determine the precise transduction rate or examine the expression of the marker on a per-cell basis. Reiser et al. used a similar vector to transduce CD34⁺ cells but did not investigate the stability of expression (34). Neither group attempted to optimize the HIV vector. Naldini et al. demonstrated that an HIV vector containing long terminal repeats (LTRs), packaging signal, marker gene, and Rev response-element could transduce resting cells (29). The HIV(VSV G) pseudotyped viral supernatant had no detectable replication-competent virus. HIV-packaging cell lines have been derived by using HIV envelope glycoprotein such that the resulting replication-defective virus had limited tropism and relatively low titer (5).

We show here that HIV(VSV G) pseudotyped viral preparations are highly efficient in transducing human CD34⁺ Thy1⁺ cells. When concentrated virus was used, the transduction rates were greater than 50% as measured by the expression of the alkaline phosphatase (AP) marker gene and close to 100% as measured by PCR of hematopoietic colonies. Transduction was dependent upon both reverse transcriptase and integrase. Transduction was optimal if the cells were exposed to cytokines for at least 48 h, but by several different criteria transduction appeared to be independent of mitosis. A series of HIV vectors were constructed, and long-term expression was greatest when there was a deletion in both Vif and Vpr in addition to Nef, Env, and Vpu. No replication-competent virus was detected in 10⁸ infectious units. These results thus extend the utility of this lentivirus gene transfer system.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Hematopoietic stem cells (HSC). Healthy human donors were primed with granulocyte colony-stimulating factor (G-CSF) for 5 days and subsequently underwent leukophoresis. Peripheral blood CD34⁺ cells were positively selected with a cell separation device (Baxter HealthCare). CD34⁺ Thy1⁺ cells were prepared by high-speed flow cytometry as described previously (37). Purified cells were frozen in 90% fetal calf serum-10% dimethyl sulfoxide or used within 24 h of isolation. The cells were typically maintained in Iscove's modified Dulbecco's medium (IMDM) supplemented with 10% fetal calf serum, 50 U of penicillin G per ml, and 50 µg of streptomycin sulfate per ml and containing 20 ng of interleukin-6 (IL-6) per ml, 20 ng of IL-3 per ml, and 100 ng of stem cell factor per ml. For clonogenic assays, the cells were placed into methylcellulose (Stem Cell Technologies) in the complete medium described above along with 2 U of erythropoietin per ml and 1 ng of granulocyte-macrophage (GM)-CSF per ml.

Plasmid vector construction. pHIV-AP was obtained from N. Landau (15). It is derived from the HIV-1 NL4-3 isolate and has a frameshift in gp160, and human placental alkaline phosphatase replaces Nef. pHIV-APenv was derived from pHIV-AP by deleting an AflIII fragment from nucleotides 6054 to 7488 (Fig. 1A). pHIV-APenvR1 was derived from pHIV-APenv by inserting 4 bp at the unique EcoRI site within Vpr at position 5743 with the Klenow fragment of Escherichia coli DNA polymerase. pHIV-APenvVifVpr was derived from pHIV-APenv by deleting the fragment from NdeI (position 5123, within Vif) to EcoRI (position 5743). pHIV-AP G-P-E-F-V- was derived from pHIV-APenvVifVpr by deleting the NsiI fragment (positions 1251 to 4381), which encompasses Gag-Pol. pHIV-AP E-F-V-T- was derived from pHIV-APenvVifVpr by inserting 4 bp at the MfeI site at position 5898, within the first exon of Tat. pHIV-AP E-F-V-R- was made by inserting 4 bp at the BamHI site (position 8465, within the second exon of Rev) of pHIV-APenvVifVpr. pHIV-AP E-F-V-R-T- was derived from pHIV-APenvVifVpr by deleting the sequence between the NdeI site (position 5123) and the AflIII site (position 7488). pHIV-PV was constructed in pCI (Promega Biotec) by using the NL4-3 fragment from BssHII (position 711) to XhoI (position 8887). A schematic of these HIV vectors is shown in Fig. 1A. pHIV-CD4 was constructed by replacing the NotI-XhoI fragment (which encompasses AP) of pHIV-APenvVifVpr with the 1.4-kb BamHI fragment encoding human CD4 of pBABEneoCD4 (a gift of J. Skowronski). pHIV-APenvVifVpr-BspEI was made by inserting 4 bp at the BspEI site at position 308 of pHIV-APenvVifVpr. These are shown in Fig. 1B. pHIV-APenv D64V was constructed by replacing the 3.1-kb ApaI-SalI fragment of pHIV-APenvVifVpr with the 3.8-kb ApaI-SalI fragment of pHIV-hygro D64V (a gift of A. Leavitt). pME VSV G, encoding the VSV G glycoprotein, and pSV A-Mo-MLV, encoding the MuLV amphotropic envelope, were gifts of K. Maruyama (DNAX) and D. Littman (New York University School of Medicine), respectively.

View larger version (18K): [in this window] [in a new window]   FIG. 1.   (A) HIV vector constructs. The parent construct was HIV-AP (reference 15 and data not shown), which is based upon NL4-3 (top). Deletions are indicated by delimited bars, and each frameshift mutation is indicated by x. In pHIV-PV, CMV designates the immediate-early cytomegalovirus promoter and polyA represents a cellular polyadenylation addition site. Genes are not precisely to scale. (B) HIV vectors with the 5' LTR BspEI site removed (for PCR analysis) and with the marker human CD4 (for FACS analysis). See the text for details.

Preparation of vector supernatants. Pseudotyped HIV supernatants were made essentially as described previously, without the addition of pcRev or butyrate (40). In brief, 293T cells were transfected with up to three plasmids by calcium phosphate coprecipitation. Supernatants were collected roughly 60 h later, centrifuged at 2,000 × g for 5 min, and subjected to titer determination with HOS cells by endpoint dilution. Before transduction of HSC, previously frozen viral stocks were concentrated 10-fold by ultrafiltration with Amicon Centriprep-10 units as specified by the manufacturer. Alternatively, supernatants were concentrated by ultracentrifugation with an SW28 rotor at 23,000 rpm at 4°C for 2 h and resuspended in 1/100 volume of IMDM by end-over-end rotation at room temperature (RT) for 3 to 6 h.

Flow cytometry and marker analysis. Cells transduced with HIV-CD4 were pelleted by microcentrifugation for 5 s and incubated for 1 h in 1:10 anti-CD4-phycoerythrin (PE) (Pharmingen) in phosphate-buffered saline-2% fetal calf serum. For DNA content measurements, washed cells were incubated at 37°C for 1 h in the presence of 0.75 mM Hoechst dye 33342 (Molecular Probes). To simultaneously measure the transduction efficiency and the number of S-phase cells by bromodeoxyuridine (BrdU) incorporation, treated cells were fixed with 0.3% formaldehyde-0.4% glutaraldehyde for 5 min at RT, washed in phosphate-buffered saline, and incubated for 1 h in anti-BrdU (Amersham) in the presence of 0.01% Tween 20 followed by fluorescein isothiocyanate-conjugated anti-mouse immunoglobulin G (FITC-IgG) (Sigma) for 1 h. Washed cells were incubated with anti-CD4-PE as described above. Analyses were carried out on a FACStar equipped with a UV laser for DNA content measurements or on a FACScan with Lysis II software (Becton Dickinson). For alkaline phosphatase staining, the cells were fixed as described above, heated at 65°C for 20 min, and incubated with 5-bromo-4-chloro-3-indolylphosphate/nitroblue tetrazolium (BCIP/NBT) along with 0.24 mg of levamisole per ml as described previously, usually for less than 2 h at RT (15). Alternatively, Vector Red (Vector Labs) replaced BCIP/NBT. The capsid antigen concentration in viral supernatants was measured with commercial reagents (Coulter).

PCR marking. Individual hematopoietic colonies were placed in 50 µl of 0.1 M KCl-10 mM Tris HCl (pH 8.3)-2.5 mM MgCl₂-0.5% Tween 20-0.5% Nonidet P-40-100 µg of proteinase K per ml and incubated overnight at 37°C. The forward and reverse oligonucleotide primers for the HIV provirus were 5'-AAGAGGCCAAATAAGGAGAGAAGAACAG-3' (NL43-171U; positions 171 to 198) and 5'-ATCTAATTCTCCCCGCTTAATACCGAC-3' (NL43-831L; positions 804 to 831), respectively, which gave rise to a 660-bp product. PCR was performed for 40 cycles at a denaturing temperature of 94°C for 30 s, an annealing temperature of 62°C for 1 min, and an extension temperature of 72°C for 1 min. The primers for -globin were 5'-ACACAACTGTGTTCACTAGC-3' and 5'-CAACTTCATCCACGTTCACC-3' and gave rise to a 112-bp product. PCR for this product was performed as described above, except that the annealing temperature was 53°C. Alu-PCR was performed as described previously (4), except that Taqpluslong (Stratagene) was used in the initial PCR step and the second PCR step was performed with 1% of the original product for 27 cycles. Products obtained with the HIV primers were size separated by horizontal agarose gel electrophoresis, transferred under alkaline conditions to Hybond N⁺ (Amersham), and probed with a ³²P-labelled DNA fragment encompassing the 5' LTR of NL4-3. Washed filters were exposed to X-ray film. For the -globin product, samples were fractionated on a 2.5% agarose gel and visualized by ethidium bromide and UV light.

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

HIV(VSV G) efficiently transduces HSC. To determine the transduction rates of pseudotyped HIV vectors, an HIV provirus with a 1.45-kb deletion in gp160 and AP replacing Nef was constructed (pHIV-APenv; Fig. 1A). Typical endpoint titers for transiently produced pHIV-APenv(VSV G) were 2.0 × 10⁷ to 3.0 × 10⁷ IU/ml on HOS cell targets. Mobilized, peripheral blood HSC, which were either CD34⁺ or CD34⁺ Thy1⁺, were obtained from volunteer donors. After isolation, the cells were placed in complete medium with cytokines and transduced overnight with ultrafiltered, 10-fold-concentrated viral stocks in the presence of 4 µg of Polybrene per ml immediately (time zero) or 24 or 48 h later. Fixed cells were assayed for AP by BCIP/NBT staining 48 h after the last time point. Although transduction rates were low at 0 and 24 h (roughly 5 and 10%, respectively), they were close to 50% at 48 h, very similar for both CD34⁺ and CD34⁺ Thy1⁺ cells (Fig. 2). Transduction was absolutely dependent upon the presence of VSV G, and it was also dependent upon reverse transcriptase, since the addition of zidovudine inhibited transduction by more than 90%. pHIV-APenv(A-Mo-MLV) typically gave two- to threefold-lower transduction rates (data not shown). This may reflect the lower titer of this viral stock (which was at most 1.0 × 10⁷ IU/ml as determined with HOS targets) or less abundant expression of the receptor for amphotropic virus envelope. To demonstrate that the expression of AP was dependent on provirus integration, an HIV construct in which one of the catalytic triad residues of IN was changed (D64V) was obtained from A. Leavitt. This virus has normal levels of reverse transcriptase activity but 1,000- to 10,000-fold-reduced titers on HOS cells (22). This mutation, when placed in the context of pHIV-APenvVifVpr, reduced HSC transduction rates to 1 to 2% of the control values.

View larger version (18K): [in this window] [in a new window]   FIG. 2.   Transduction increases if HSC are exposed to cytokines for 2 days. CD34+ or CD34+ Thy1+ cells were placed into cytokine-containing medium, and at 0, 24, and 48 h ultrafiltered HIV-APenv(VSV G) was added for an overnight incubation. For the bald virus, the VSV G expression plasmid was omitted from the original transfection. To inhibit reverse transcriptase, zidovudine (AZT) was added to a final concentration of 0.5 mM. All cell samples were fixed at 96 h, developed with BCIP/NBT, and scored visually for a brownish-purple color change. At least 100 cells were counted for each determination.

Replication-competent virus is undetectable. It is conceivable that a double, nonhomologous crossover event could occur during the transient transfection of the 293T cells with the VSV G envelope and replication-defective HIV provirus plasmids, resulting in replication-competent HIV with a wide host range and capable of spreading infection. To test this remote possibility, 5 × 10⁶ HOS cells were transduced with 5.0 ml of pHIV-APenvVifVpr(VSV G) with a titer of 3 × 10⁷ IU/ml such that more than 95% of the targets were transduced. A 10-ml volume of supernatant from these transduced targets was used to transduce 5 × 10⁶ naive HOS cells, and the resulting titer was 10³ IU/ml. This process was performed iteratively, and by the fourth round, the titer was undetectable. In addition, the >95% transduced HOS cells described above were serially passaged 1:4 or 1:8 every 3 to 5 days. Each week, 1 ml of supernatant was used to transduce naive HOS cells, with a resulting titer of <10 IU/ml, which did not change over a 5-week period. In addition, the few HOS cells that were transduced were passaged as described above, and the number of positively staining cells remained stable over an 8-week period. The lack of viral spread suggests that the few events observed were a result of non-envelope-mediated viral entry (i.e., endocytosis of envelope-negative, replication-defective virus).

Transduction of HSC by HIV is cell cycle independent. One of the principal reasons for developing lentivirus vectors is to exploit their ability to transduce nondividing targets. However, the results shown in Fig. 2 suggested that transduction may be cell cycle dependent. To address this question, purified HSC were placed in culture with cytokines and incubated 24 h later with increasing concentrations of aphidocolin (S-phase inhibitor). The cells were then transduced overnight in the presence of aphidocolin, refed, and stained for AP activity 24 h later. No clear inhibition of transduction was observed, as measured by BCIP/NBT or Vector Red staining (Fig. 3). To further explore this issue, a replication-defective HIV was constructed with CD4 as a marker for fluorescence-activated cell sorter (FACS) analysis (Fig. 1B). CD34⁺ cells were transduced with HIV-CD4(VSV G) overnight, and stained for DNA content with Hoechst dye 33342 and CD4 expression 48 h later. As shown in Fig. 4, the DNA content of transduced cells was almost identical to that of nontransduced cells (and the population as a whole). This is consistent with transduction being independent of the cell cycle. However, in this experiment, we could not determine the precise stage the cells were in when they were transduced. To do this, CD34⁺ cells were labelled with BrdU at the same time as they were transduced overnight with HIV-CD4(VSV G). At 48 h later, fixed cells were labelled with anti-BrdU and then with FITC-IgG. Washed cells were then incubated with anti-CD4-PE and analyzed by flow cytometry. The results in Fig. 5 indicate that labelling by BrdU and transduction were independent events. Similar results were obtained with a 4-h labelling-transduction period, although the proportion of the cells which were labelled or transduced was lower.

View larger version (13K): [in this window] [in a new window]   FIG. 3.   Aphidocolin does not inhibit transduction. Increasing amounts of aphidocolin (Sigma) were added to CD34+ Thy1+ cells 1 day before overnight transduction with concentrated HIV-APenv(VSV G) and left in the medium until fixation was carried out. Transduction was scored by BCIP/NBT or Vector Red staining as specified by the manufacturer. Vector Red caused positive cells to be bright red, and they were easily visible by light microscopy.

View larger version (32K): [in this window] [in a new window]   FIG. 4.   Transduced cells have a DNA content profile similar to that of untransduced cells. CD34+ cells were transduced overnight with ultracentrifuge-concentrated bald HIV-CD4 (left) or HIV-CD4(VSV G) (right). Two days later, the cells were analyzed by flow cytometry on a FACStar equipped with a UV laser. For CD4 measurements, the primary antibody was mouse anti-CD4-PE used at a 1:10 dilution. For DNA content measurement, the cells were incubated with 0.75 mM Hoechst dye 33342 (Molecular Probes) for 1 h at 37°C. Only live cells were quantified; they were gated by exclusion of the dye propidium iodide. N.D., not determined.

View larger version (16K): [in this window] [in a new window]   FIG. 5.   Transduction is independent of the S phase. CD34+ cells were transduced overnight with ultracentrifuge-concentrated bald HIV-CD4 (mock) or HIV-CD4(VSV G) and at the same time incubated in the presence or absence of 30 µg of BrdU per ml. Two days later, the cells were fixed and incubated with mouse anti-BrdU followed by FITC-IgG. After further washing, the cells were incubated with mouse anti-CD4-PE (1:10 dilution) and analyzed by flow cytometry on a FACScan. Similar results were obtained in a 4-h experiment. BrdU-positive cells represent those that were in the S phase.

Stable expression is dependent upon deletion of Vif and Vpr. To determine the stability of expression of the introduced marker gene, CD34⁺ cells were transduced overnight with pHIV-APenvR1(VSV G) or bald virus. This HIV construct has a frameshift in Vpr at the EcoRI site and functionally behaves as a Vpr virus in established cell lines in that transduced cells do not undergo cell cycle arrest but instead have normal growth properties. A 50% transduction rate was confirmed 48 h later, and the cells were plated into methylcellulose in the presence of IL-3, IL-6, stem cell factor, GM-CSF, and erythropoeitin to allow colony differentiation and proliferation. Two weeks later, colony types were counted and stained for AP activity. To our surprise, the cells transduced with VSV G-pseudotyped virus produced roughly 50% fewer colonies of each type than did the cells transduced with otherwise identical bald virus (Fig. 6). Very few of the colonies stained positive for AP, and those colonies were minute (fewer than 50 cells). This result, reproduced several times, suggested that there was a cytotoxic or cytostatic gene product present within pHIV-APenvR1 which did not allow for expansion of transduced cells. We then made a series of deletion and frameshift vector constructs (Fig. 1A) and tested them for initial transduction efficiency and stability of expression by bulk culture and by colony formation in methylcellulose. As shown in Table 1, each construct, pseudotyped with VSV G, had a different titer on HOS cells and a different initial transduction efficiency on CD34⁺ cells, with pHIV-APenvVifVpr being superior in this respect. AP expression was monitored in bulk culture for 4 weeks. For each of the constructs in Fig. 1A, there was a decay in the proportion of cells with detectable expression (Fig. 7). This decay was most pronounced and significant for pHIV-APenvR1 (compared to pHIV-APenvVifVpr), consistent with the results described above. On day 3, a fraction of the transduced cells were plated into methylcellulose and examined for AP expression after 3 weeks. The percent AP-positive colonies mirrored what was observed in bulk culture (data not shown), although there was a subpopulation of sectored colonies. These results suggest that either Vif or the amino terminus of Vpr is cytotoxic or cytostatic to CD34⁺ cells.

View larger version (26K): [in this window] [in a new window]   FIG. 6.   Transduction by HIV-APenvR1(VSV G) results in cytotoxicity. CD34+ Thy1+ cells were transduced overnight with ultrafiltered bald or HIV-APenvR1(VSV G). Three days later, the cells were plated into methylcellulose. Colony types and AP staining were determined 2 weeks later. GM, granulocyte-macrophage; E, erythroid; mixed, mixed cell types. Only a few AP+ colonies were observed, and these were small compared to AP colonies. This experiment was repeated twice more with CD34+ cells, and similar results were obtained.

View larger version (17K): [in this window] [in a new window]   FIG. 7.   Expression of AP marker declines over time. CD34+ cells were transduced in triplicate overnight with ultracentrifuge-concentrated VSV G pseudotyped HIV vectors (Table 1). Note that the viruses HIV-AP G-P-E-F-V-, HIV-AP E-F-V-T-, HIV-AP E-F-V-R-, and HIV-AP E-F-V-R-T- were produced by cotransfection with pHIV-PV (Fig. 1A) and pME VSV G. After transduction, the cells were either plated out into methylcellulose and 3 weeks later stained for AP with BCIP/NBT or maintained in bulk culture in the presence of IL-3, IL-6, and stem cell factor and periodically stained for AP. No staining was observed for cells infected with bald HIV-APenvVifVpr (data not shown). For each construct, the percentage of AP+ colonies observed in methylcellulose was similar to what was seen in bulk culture. *, P < 0.0001 compared to HIV-APenvVifVpr; **, P > 0.05 compared to HIV-AP E-F-V-R- but P < 0.0001 compared to HIV-APenvVifVpr; §, P < 0.0001 for both HIV-APR1 and HIV-AP E-F-V-R-T- compared to HIV-APenv Vif Vpr (all results obtained by two-way balanced analysis of variance). Error bars have been omitted for clarity.

View larger version (48K): [in this window] [in a new window]   FIG. 8.   PCR assays for vector proviruses in MC colonies. CD34+ cells were transduced overnight with ultracentrifuge-concentrated HIV-APenvVifVpr-BspE1(VSV G) or the bald control virus stock, treated with DNase 1 at 100 µg/ml in the presence of 5 mM MgCl2 for 24 h, and then plated into methylcellulose. (A) Colonies contain fully replicated forms of HIV provirus. In this experiment, 42% of the cells initially stained positive for AP. Three weeks later, DNA was prepared from individual colonies and PCRs were carried out with primers NL43-171U and NL43-831L as described in Materials and Methods. At that time, only 24% of the colonies stained positive for AP. A portion of each PCR product was digested with BspEI, size fractionated on a 1.2% agarose gel, transferred to Hybond N+ (Amersham), and hybridized to a 32P-labelled DNA probe encompassing the HIV-1 LTR. The filter was washed with 0.2× SSC (1× SSC is 0.15 M NaCl plus 0.015 M sodium citrate)-0.1% sodium dodecyl sulfate at 65°C for a total of 30 min and exposed to X-ray film for 20 min. First 12 lanes, bald virus; last 12 lanes, HIV-APenvVifVpr-BspE1(VSV G). The expected product size prior to BspEI digestion is 660 bp; after digestion, it is 523 bp. None of the first 12 and 10 of the last 12 reactions were judged to be positive. When -globin control primers were used, only one sample was negative. (B) Colonies contain integrated forms of HIV provirus. In this experiment, the initial transduction rate was 65% and at 3 weeks 24% of the colonies stained positive for AP. DNA was prepared from individual colonies, and all 24 were positive when the -globin control primers were used. None of the mock-transduced and all 12 of the HIV-APenvVifVpr-BspE1(VSV G)-transduced colonies were positive by PCR with the primers used in panel A. Samples were then subjected to Alu-PCR as described previously (4), and the products were size fractionated on a 1.5% agarose gel prestained with ethidium bromide. None of the mock-transduced and 10 of 12 HIV-APenvVifVpr-BspE1(VSV G)-transduced colonies gave the expected size of product, as indicated.

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

We demonstrate efficient transduction of purified human HSC by HIV vectors pseudotyped with VSV G. Titers of transiently produced pseudotyped HIV reported here are equivalent to or greater than those previously reported (1, 28, 29, 34), which may reflect the nature of the enzymatic marker AP. In studies with CD34⁺ cells, the HIV proviral construct simply had a deletion in Env, with the marker gene inserted within Env or Nef, and the stability of marker gene expression was not examined. In the results reported here, since more of the HIV genome was deleted from the vector, the titer was reduced accordingly, which may be due in part to inefficient cotransfection of the 293T producer cells, with the plasmid providing complementing functions in trans. Alternatively, packaging or reverse transcription of deleted vectors may not be optimal. The transduction rates we observed for HSC exceeded those of other currently used retroviruses (6, 12, 20, 21, 27) and were roughly equivalent to those of other viral vector systems in use (3, 7, 13, 26, 30, 45). These are likely to be true transduction rates, since they were dependent upon both reverse transcriptase and integrase. We were surprised that the titers of the Tat HIV construct (pHIV-AP E-F-V-T-) were reduced 30- to 50-fold on HOS cell targets whereas CD34⁺ transduction rates were diminished only 2- to 3-fold. The Tat virus was prepared by cotransfection of the pHIV-PV construct, which has an intact Tat gene, so that production of the virus would presumably not be limiting. It is possible that the AP assay is more sensitive and nonlinear than measurements of transcript abundance. Readthrough transcription from bona fide or fortuitous promoters in cellular flanking sequences followed by conventional splicing would also yield an AP-positive cell. The titers of the construct lacking functional Tat and Rev were reduced 100-fold on HOS cells and gave very poor transduction rates on HSC.

The transduction rates in our experiments were reproducibly higher if HSC were preactivated with cytokines for 48 h. However, this high transduction rate was independent of mitosis in that aphidocolin, an S-phase inhibitor, had no consistent effect on the transduction efficiency. In addition, transduced cells had nearly the same DNA content profile as untransduced cells, with an insignificant bias toward the S, G₂, and M phases. Furthermore, transduction by HIV(VSV G) was independent of the S phase, as indicated by the BrdU labelling experiment. These results, taken together, are consistent with previous findings that HIV can infect nondividing cells. It is not clear why we observed a requirement for the cells to be cultured in the presence of cytokines for a few days for maximal transduction rates, which is at odds with the results discussed above (34). However, the CD34⁺ cells were prepared and transduced in a slightly dissimilar manner, and different markers and detection systems were used. More than 99% of freshly isolated HSC are in the G₀/G₁ (presumably G₀) phase as measured by DNA content analysis (34, 42). HSC require cytokines both to prevent apoptosis and to enter G₁. The results presented here may be reconciled with the findings that efficient, complete transduction by HIV probably requires the target cell to be transcriptionally active and at least in G₁ (out of G₀) (38, 39, 41). HIV can enter G₀ CD4⁺ T cells and macrophages, but reverse transcription and expression of viral gene products are limited (41, 47). Importantly, these cells need not traverse mitosis for nuclear entry and expression of viral genes to take place. Thus, if transduction is measured by transgene expression (as it was in most cases here), a requirement for an activated state and consequent transcription but not mitosis is observed.

We also demonstrate that expression of the transgene wanes over the course of several weeks, so that an initial transduction rate measured by expression of more than 50% (as an example) declines to a proportion of 25% at 3 weeks. It is not known whether expression diminishes further over more extended times. However, transduction measured by PCR marking remained high. It is uncertain which features of the virus (or host cell) cause extinction of viral gene expression, but this phenomenon is generally observed for the retroviruses and remains problematic for their use as gene transfer agents. An unexpected finding presented here is that HIV-1 contains a gene product which is selectively cytotoxic to HSC. This product is not Tat or the carboxy terminus of Vpr, which have been shown in other cell types to cause apoptosis (24, 32, 46) and G₂/M arrest (14, 17, 25, 33), respectively. Based on the series of deletion constructs, the cytotoxic product is either Vif or the amino terminus of Vpr. Other experiments suggest that this property maps within Vif, not Vpr (data not shown). Although Vif is highly conserved among different HIV-1 isolates and is present in other lentiviruses, its role in the viral life cycle remains poorly defined. Vif is required for viral replication in primary cell types and is critical for proviral DNA synthesis in selected target cells (44). The cytotoxicity of certain HIV-1 isolates has been mapped to the Vif gene product (36), with the cytopathic effect being manifested as loss of cell viability and giant cell formation. Others have reported that cell clones that survive the initial cytopathic effect harbor HIV species which have individual mutations in accessory gene products, including Vif (18, 19). However, toxic effects of Vif are not consistently seen (44), and we do not observe the cytotoxic effect of Vif in established human and mouse cell lines. For the HSC transduced with Vif⁺ HIV, multinucleated giant cells were not observed, but we do not yet know the cause of the loss of cell viability. These results do suggest, however, that deletion of Vif is a requisite for maintenance of the transduced HSC population and hence for expression of the transgene.

We have yet to detect replication-competent virus present in the transient-transfection viral supernatants. This has also been true for similar HIV vector systems, which have larger deletions of the provirus (5, 28, 29, 31). It remains possible that replication-competent virus exists at a level of 10⁹ or 10¹⁰ (representing 30 or 300 ml of viral supernatant, respectively) and that the assay used here is not sensitive enough to detect these rare occurrences. It is also conceivable that a target cell could express an endogenous envelope so that cells transduced with pHIV-APenv would produce virus of altered tropism but would still be replication defective. It is thus most desirable to use a vector such as pHIV-AP G-P-E-F-V- or the previously described transfer vector (29), which has a minimum of residual HIV sequence. Packaging cell lines with HIV core proteins and HIV envelopes have been described, but the host range of the replication-defective virus was more restricted and the titers were reduced at least 1,000-fold compared to HIV(VSV G) pseudotypes (5, 31). Transient production of viral supernatants is clearly advantageous for vector development, but it will be desirable ultimately to generate suitable HIV packaging cell lines of wide host range in a bioreactor system, as has been demonstrated for MuLV (8).