Transduction of Human Immunodeficiency Virus Type 1 Vectors Lacking Encapsidation and Dimerization Signals

Abstract

The encapsidation signal (Ψ) and the nested dimerization initiation site are important for efficient packaging of human immunodeficiency virus type 1 (HIV-1) genomic RNA dimers. Consequently, these signals are included in all HIV-1 vectors. Here, we provide evidence demonstrating that these elements in such vectors are not absolutely required for vector transduction. In single-cycle infection assays, vectors with Ψ deleted (ΔΨ) were transduced with only a two- to fivefold reduction compared to the wild type. The transduction of ΔΨ showed typical products of reverse transcription and vector integration; however, in vitro and in vivo dimerization assays demonstrated the lack of normal dimerization of the ΔΨ vector. The reduction in transduction reflected a similar reduction in packaging. Nevertheless, a relatively high specificity of packaging was retained, as the ΔΨ vector was encapsidated at a level 4 orders of magnitude higher than that for overexpressed, nonretroviral cellular mRNA and 15 orders of magnitude higher than that for a murine leukemia virus (MLV)-based vector, all containing the same reporter gene, suggesting a Ψ-independent mechanism of packaging. The fact that HIV-1 and MLV vectors were encapsidated with a much higher level of efficiency than the cellular RNA suggests that the genomic RNAs of different retroviruses share common features and/or pathways that target them to encapsidation. Overall, these results formally demonstrate that packaging and dimerization signals are not required for the early stages of infection and can be deleted without risking a total loss of vector transduction. Deletion of these signals should enhance the safety of these vectors.

Interactions between retroviral genomic RNA and the Gag structural protein are thought to mediate efficient incorporation of the virus genome into the assembled virion. These interactions include specific binding between the nucleocapsid (NC) domain of the Gag precursor and the core encapsidation signal (E or Ψ) that resides at the 5′ portion of the RNA genome (reviewed in references 11 and 24). The major packaging signal of HIV-1 is a structured RNA (8, 31, 33, 63) consisting of four adjacent stem-loops (12, 16, 18, 45-47) located downstream of the primer binding site and extending into the 5′ portion of the gag gene. Each of the four stem-loops (SL1 to SL4) appears to have substantial ability to bind HIV Gag, and the same is true for multiple stem-loops consisting of two, three, or four of these stem-loops in tandem (2, 4, 12, 16, 23, 74). The affinity of this binding is estimated to range between a K_d of ∼20 to ∼400 nM, depending on the type of assay, use of the Gag precursor versus the NC portion, and the different RNA molecules tested (3, 12, 16, 23, 69). Other studies revealed that, in HIV-1, SL1 and SL3 are important for the selective encapsidation of the genomic RNA, as mutations in these structures hamper the discrimination between spliced and unspliced viral RNAs (18, 32, 46). In addition, stabilization of the Ψ structure has been proposed to be an indirect role of SL4 in Gag-Ψ interactions (3).

The packaging signal harbors yet another functional sequence, the dimerization initiation site (DIS). This sequence promotes the dimerization of two genomic RNA molecules, the form found packaged in all members of the Retroviridae family (29, 30, 56). In HIV-1, this signal is located in SL1, where according to the kissing loops model, a palindromic sequence in the loop of SL1 of one genomic RNA molecule anneals to the same sequence found in a second genomic RNA molecule (10, 39, 57, 70, 73). The annealing reaction between these two genomes extends to sequences adjacent to DIS to form a dimer linkage structure, resulting in a stable, noncovalent association between the two genomes near their 5′ ends. Although the purpose of this dimerization is not clear (61), its conservation in all retroviruses suggests that it has an important role in the virus life cycle (29). Dimerization of RNA molecules containing the DIS/dimer linkage structure sequence can occur in vitro (for examples, see references 44, 57, and 70), and the Gag and NC proteins assist the formation of a stable dimer (20, 21, 26, 28). In addition, in vivo chemical modification assays demonstrated that the HIV-1 genome could already dimerize in the cytoplasms of infected cells (55). These findings and the fact that the DIS and Ψ signals overlap have raised the possibility that the dimerization of the two genomes may assist in their packaging. Indeed, D'Souza and Summers have recently suggested that the dimerization of the two genomes of the Moloney murine leukemia virus (MoMLV) promotes the exposure of high-affinity NC binding sites in the packaging signal, leading to efficient encapsidation of the dimeric RNA genomes (24, 25). It has been suggested that, for HIV-1, a riboswitch, which is a transition between alternative conformations of the Ψ and DIS signals, regulates RNA dimerization and packaging (9, 54).

The vast amount of data describing the importance of the Ψ/DIS signal to the efficient replication of HIV-1 has led to the inclusion of these cis-acting elements in all HIV-1-based vectors constructed to this date (37, 51, 52). These vectors are attractive vehicles for gene delivery because of their efficient infection of human tissues, including nondividing cells (52). Yet the high pathogenicity of HIV-1 imposes safety issues on the design and use of vectors that derive from this pathogen (38). To date, this issue has been addressed mainly by minimizing the presence of viral sequences in the vector and the separation of trans- and cis-acting elements into individual plasmids. These modifications reduce the risk of unwanted recombination and generate particles that are restricted to a single cycle of infection. Here we show that HIV-1-based vectors with deletions of SL1 to SL4 can still be efficiently packaged and transduced, albeit with no apparent normal dimerization. Although the overexpression of these vectors in the producer cells may contribute to their encapsidation, these genomic RNA molecules are packaged at a much higher level of efficiency than overexpressed cellular mRNAs harboring the same reporter gene, suggesting the presence of a Ψ-independent mechanism of packaging. Thus, deletion of Ψ/DIS should allow for safer use of these vectors, minimizing the formation of replication-competent virus through unwanted recombination events.

MATERIALS AND METHODS

Mammalian expression plasmids.

pCMVΔR8.2 (51) encodes HIV-1 proteins except the envelope and the Vpu; pMD.G encodes the vesicular stomatitis virus G envelope protein (VSV.G) (51); and pHR′CMV-lacZ (52), pHR′CMV-GFP (49), and pHR′CMV-IRES-neo (see below) encode an HIV-1-derived retroviral vector carrying either β-galactosidase, the enhanced green fluorescence protein (EGFP), or the neomycin resistance (neo) gene, respectively. These plasmids were generously provided by I. Verma (Salk Institute). pEGFP-N3 (Clontech) is a mammalian expression plasmid encoding the EGFP marker. pQXIP-EGFP (designated Mvec) encodes a MoMLV-derived vector carrying the EGFP reporter gene (42).

Mutagenesis.

Deletion of the Ψ signal (including SL1 to SL4) (Fig. 1A) in the HIV vector pHR′CMV-GFP to create pHR′CMV.GFPΔΨ was achieved by PCR, using the following primers: HIV-SL4 forw and HIV-SL1 rev (Table 1), which bind from the 3′ stem sequence of SL4 and the 5′ stem sequence of SL1, respectively. The two primers share a PmeI restriction site. In a PCR using these primers and a Pfu polymerase (Stratagene), linear copies of plasmid pHR′CMV-GFP, excluding 105 base pairs (bp) of the original Ψ sequence, were synthesized. The PCR product was initially digested with DpnI to selectively remove the template of the parent plasmid, pHR′CMV-GFP, and then it was digested with PmeI, self-ligated with T4 ligase (MBI Fermentas), and transformed into Escherichia coli (JM109). Clones bearing the ΔΨ mutation were verified by sequencing. To minimize the effects of PCR-generated mutations in the plasmid backbone, an ApaI-NotI fragment (2.16 kilobases [kb]) that includes the ΔΨ mutation was subcloned into the corresponding site in pHR′CMV-GFP and pHR′CMV-lacZ to generate pHR′CMV-GFPΔΨ and pHR′CMV-lacZΔΨ, respectively. In addition, using the unique BamHI and XhoI sites, the lacZ cassette in both pHR′CMV-lacZ and pHR′CMV-lacZΔΨ was replaced by a cassette carrying the internal ribosome entry site (IRES)-neomycin resistance gene derived from pQCXIN (Clontech), hence generating pHR′CMV-IRES-neo and pHR′CMV-IRES-neoΔΨ, respectively. The ΔΨ mutagenesis procedure introduced 12 nucleotides, which included the splice donor (SD) and a unique PmeI site replacing a unique BssHII site in SL2, to the site of the deletion (Fig. 1B).

FIG. 1.

Schematic diagrams of HIV-1 vectors used in this study. (A) Schematic diagram of the pHR′ vector and expanded view of its core encapsidation signal. Shown are the 5′ and 3′LTRs (gray boxes with R sequence highlighted in black), the truncated gag open reading frame (boxed with a wavy line), the Rev-responsive elements (RRE), and the CMV internal promoter driving the reporter gene (in this study, gfp or lacZ). The drawing is not to scale. The expanded view of the encapsidation signal shows the four stem-loops (SL1 to SL4). The palindromic sequence of the DIS, which also forms a BssHII restriction site, is boxed. The SD site is represented by an arrowhead. Shaded letters represent the start codon of the gag open reading frame. The sequence between the solid arrows was deleted to make the ΔΨ mutation. (B) Representation of the sequence replacing SL1 to SL4. In the ΔΨ mutant, a PmeI restriction site was introduced together with the SD site.

TABLE 1.

List of primers with their applications

Primer	Sequence (5′→3′)	Application
HIV-SL4 forw	ATGTGTTTAAACTGGTAGTATTAAGCGGGGGAG	ΔΨ mutation
HIV-SL1 rev	ATCATGTTTAAACCCGAGTCCTGCGTCGAG	ΔΨ mutation/alu PCR
2LTRcircle-F	AACTAGGGAACCCACTGCTTAAG	LTR circle PCR
2LTRcircle-R	TCCACAGATCAAGGATATCTTGTC	LTR circle PCR
T7HIV5′R	TACGACTCACTATAGGGTCTCTCTGGTTAG	Genomic PCR
Gag9	TTCTGATCCTGTCTGAAGGG	Genomic PCR
−sssDNA U5	CTGCTAGAGATTTTCCACACTGAC	alu PCR
Alu	GCCTCCCAAAGTGCTGGGATTACAG	alu PCR
T7HIV-TAR(F)	CGATCCTAATACGACTCACTATAGGGTCTCTCTGGTTAGAC	In vitro transcription
GFP-Nhe1(F)	GTTTAAACGGTGCTAGCATGGTGAGCAAGGGCGAGG	RT-PCR
GFP-515R	ATGTTGTGGCGGATCTTGAAG	RT-PCR
GFP-F	AAGCTGACCCTGAAGTTCATCTG	qPCR
GFP-R	TTGAAGAAGTCGTGCTGCTTCAT	qPCR
GFP-MGB probe	6FAM-ACCGGCAAGCTGC-MGBNFQ	qPCR

Transfection of 293T cells.

293T cells were maintained in Dulbecco's modified Eagle medium (DMEM) supplemented with 10% fetal calf serum and l-glutamine (DMEM-10; Biological Industries, Israel) at 37°C and 5% CO₂. The cells were transfected, using calcium phosphate (5), with 10 μg of pHR′ vectors, pEGFP-N3, or pQCXIP-EGFP, in addition to 7.5 μg pCMVΔR8.2 and 2.5 μg plasmid pMD.G DNA (per 60-mm plate). Two days posttransfection, the cells and culture media were harvested and expression of the encoded proteins was analyzed. The transfected cells were used for either fluorescence-activated cell sorting (FACS) analysis or RNA or protein extraction. HEPES (N-2-hydroxyethylpiperazine-NÍ-2-ethanesulfonic acid) was added to the culture supernatants to a final concentration of 0.05 M, and the mixture was filtered through a 0.45-μm filter. The virion-containing media were subsequently used for infection, RNA extraction, protein extraction, and/or exogenous reverse transcriptase (RT) assays. Cell lysate preparations and virion-like particle (VLPs) purifications were carried out as detailed by Melamed et al. (48).

Transduction assays.

To quantify the influence of the ΔΨ mutation on vector transduction, we utilized single-cycle infectivity assays (5) and measured the transduction of either a gfp- or lacZ-containing vector. Virion-containing media were diluted to either 1:2 or 1:3 (as indicated) with fresh DMEM-10 containing polybrene (hexadimethrine bromide; final concentration, 8 μg/ml; Sigma) and incubated with naïve 293T cells for 2 h, after which fresh DMEM-10 was added and the cells were incubated for a further 48 to 64 h. Following infection, the cells were harvested and processed for FACS analysis or assayed for β-galactosidase activity. In a subset of experiments, cells were treated with the RT inhibitor nevirapine (12 μM) for the duration of the infection. Transduction of the vectors expressing the neomycin resistance gene was assayed by infection of HeLa cells and subsequent selection of these cultures with Geneticin (G418 sulfate, 1 mg/ml; Life Technologies) for 2 weeks.

FACS.

Cells either transfected or infected with a GFP vector were fixed with 2% paraformaldehyde and analyzed by FACS for GFP fluorescence, using a FACSort apparatus (Becton Dickinson). Mock-transfected cells (cells transfected without the addition of plasmid DNA) were used to determine the background autofluorescence levels of GFP-negative cells. Total fluorescence of the GFP-positive cell population was calculated by multiplying the percentage of GFP-positive cells with their mean fluorescence intensity. Normalized infectivity was calculated by dividing the total fluorescence of the infected cells by the total fluorescence of the transfected cells (42, 48). The multiplicity of infection was calculated from the percentage of uninfected cells (GFP-negative cells) (27).

Exogenous RT assay.

To estimate the virion content in the conditioned media, RT activity was determined in an exogenous RT assay as previously described (72). Duplicate samples of the reaction mixture (5 μl) were spotted on DE81 paper, and the membrane was exposed to a phosphor screen. The intensity of the signal was read by a phosphor imager (FLA2000), quantified by using the TINA software program, and reported in arbitrary pixel units (5).

β-Galactosidase assay.

Cells were resuspended in 0.25 M Tris Cl (pH 7.8) and disrupted by three freeze-thaw cycles. The suspension was centrifuged at 12,000 × g for 5 min at 4°C. The cleared extracts (at several dilutions) were then used in a β-galactosidase assay as described by Sambrook et al. (66). The assay was modified for a microtiter plate by reduction of all the volumes to 1/3. β-Galactosidase activity was determined over the linear portion of readings for optical density at 420 nm and standardized to the total protein content of the lysates (as determined by a Bradford assay). The background activity of the mock-transfected cells was subtracted from that of the test samples from each experiment. For infected cells, β-galactosidase activity was normalized to the exogenous RT activity in the media that was used to infect the cells. The data are reported in arbitrary units (see above).

PCR to detect viral DNA sequences in infected cells.

To detect long terminal repeat (LTR)-LTR junction forms, low-molecular-weight DNA was extracted from infected cells at 16 h postinfection, using the Hirt extraction method (6, 35). LTR-LTR junctions were amplified, using the primers 2LTRcircle-F and 2LTRcircle-R (Table 1).

To demonstrate the presence and integration of wild-type (wt) Ψ and ΔΨ sequences in infected cells, genomic DNA was purified from infected cells (using an EZ RNA kit according to the manufacturer's instructions; Biological Industries). The genomic DNA was digested with DpnI to selectively remove traces of contaminating plasmid DNA prior to PCR. To demonstrate the presence of the wt Ψ and the ΔΨ mutation in vector DNA, a PCR fragment was amplified from genomic DNA, using primers spanning the Ψ region. Two primers were used, T7HIV5′R and Gag9 (Table 1), at a final concentration of 0.4 μM with 0.5 U ExTaq polymerase (Takara) in a 50-μl reaction mixture. The PCR products were digested with BssHII (unique to wt Ψ) and PmeI (unique to ΔΨ) to further distinguish between the wt and ΔΨ sequences.

To detect the integration of the wt and ΔΨ vectors into the genomic DNA of infected cells, we applied Alu PCR analysis (19). Fragments of cellular genomic DNA were amplified, using Taq polymerase (Bioline), a reverse primer specific to the vector sequences (HIVSL1 Rev), and a primer corresponding to Alu sequences (Table 1). Amplification was followed by nested PCR, using the Alu primer and an internal primer (−sssDNA U5) (Table 1). The PCR products were then separated on a 1% agarose gel, visualized by ethidium bromide staining, and subsequently transferred to a Hybond-N membrane (Amersham) for hybridization. The membrane was hybridized with a vector-specific antisense riboprobe harboring the U3 and R sequences (U3-R-gfp) as described below (see paragraph headed “In vivo dimerization”).

RT-PCR.

Total cellular RNA was extracted from cells harvested 48 h after transfection with an EZ-RNA kit (Biological Industries) according to the manufacturer's instructions. Purification of virion RNA from the culture supernatants (140 μl) was performed by using a QIAamp viral RNA mini kit (QIAGEN), and the RNA was resuspended in 60 μl of diethyl pyrocarbonate-treated water. All RNA samples were then treated with RNase-free DNase I according to the manufacturer's instructions (DNase treatment and removal kit; Ambion). First-strand cDNA synthesis was performed in a total volume of 25 μl, using avian myeloblastosis virus RT (15 U) and random hexamers (0.5 μg) in the presence of avian myeloblastosis virus RT buffer, RNAsin (25 U RNase inhibitor), and deoxynucleoside triphosphates (1 mM each). All the above reagents were purchased from Promega (Madison, WI). The RT reaction was carried out at 42°C for 1 h, and 2 μl of the RT reaction mixtures and serial dilutions of the samples were amplified by PCR, using 2× Taq mix purple (Lambda Biotech) with 0.5 μM of each of the GFP-derived primers [GFP-Nhe1(F) and GFP-515R; Table 1]. The PCR samples (up to 50%) were electrophoresed in a 1.5% agarose gel.

Real-time qPCR.

Primers and a minor groove binding protein (MGB) probe (Applied Biosystems) (Table 1) specific to GFP were used for real-time quantitative PCR (qPCR). The MGB probe was labeled with the fluorescent dye 6-carboxyfluorescein at the 5′ end and a nonfluorescent quencher at the 3′ end. The 20-μl reaction mixtures consisted of 10 μl of 2× TaqMan universal PCR master mix, 0.5 μM of each primer, 0.25 μM of MGB probe, and 2 μl of either a sample or the standard: the cDNA samples were diluted 1:10 for qPCR analysis, and the standard used was log dilutions of pHR′CMV-GFP ranging from 1 ng to 1 fg. Real-time qPCR was carried out in an ABI Prism 7000, and the data were analyzed by the corresponding software (Applied Biosystems).

In vivo dimerization assay.

293T cells in two 90-mm plates were transfected with wt or ΔΨ vectors and helper plasmids. Two days posttransfection, 20 ml of media from each transfection mixture was harvested and treated with DNase I (50 U/ml at 37°C for 1 h). The virions were then purified by ultracentrifugation through a 25% sucrose gradient at 25,000 rpm for 2 h. A dimerization assay was carried out as described by Fu et al. (28). Briefly, purified virions were resuspended in 500 μl of lysis buffer consisting of 50 mM Tris (pH 7.4), 10 mM EDTA, 1% sodium dodecyl sulfate (SDS), 100 mM NaCl, 50 μg/ml of yeast tRNA, and 100 μg/ml of proteinase K, and the virions were disrupted at 37°C for 30 min. The RNA was extensively purified by phenol-chloroform extraction (500 μl three times) and precipitated by sodium acetate and ethanol. The RNA was resuspended in 100 μl of RNA resuspension buffer (10 mM Tris [pH 7.5], 1 mM EDTA, 1% SDS, 50 mM NaCl). Aliquots (20 μl) of the vector RNA were incubated for 10 min at the following temperatures: 25, 35, 40, 44, 48, 56 and 60 °C. The samples were immediately placed on ice, mixed with 6× RNA loading buffer (Fermentas), and subjected to Northern analysis after being electrophoresed in a 1.2% native agarose gel in 1× TBE. The gel was incubated in 6% formaldehyde at 65°C for 30 min before the RNA was transferred to a nitrocellulose Hybond-N membrane (Amersham) overnight in 20× SSC (1× SSC is 0.15 M NaCl plus 0.015 M sodium citrate). The RNA was UV cross-linked to the membrane, which was blocked for 1 h at 42°C with hybridization solution (5× SSC, 50% deionized formamide, 100 μg/ml salmon sperm DNA, 1.5% SDS). The membrane was hybridized at 50°C overnight, with a 100 ng/ml vector-specific digoxigenin (DIG)-labeled antisense RNA probe. This riboprobe (U3-R-gfp) was transcribed in vitro using XbaI-digested plasmid pGP3 as a template. This plasmid contains an HIV-1 vector sequence with an inverted T7 promoter attached downstream of the U3 R sequence. The RNA runoff generated from this template contains the antisense sequences of R, U3, and gfp. The transcribed riboprobe was labeled, using T7 RNA polymerase and DIG-labeled UTP according to the manufacturer's instructions (Roche Diagnostics GmbH). The membrane was washed twice with 2× SSC-0.1% SDS buffer and twice with 0.1 SSC-0.1% SDS buffer for 30 min each, blocked with 5% skim milk in TBST (1 mM Tris, 0.1 M NaCl, 0.1% Tween 20 [pH 7.5]) and incubated with a peroxidase-conjugated anti-DIG antibody (α-DIG-POD; 1:1,000; Roche Diagnostics GmbH) for 1 h at RT. The excess antibody was washed off in TBST (four times) and incubated with chemiluminescence solution containing 150 mM Tris (pH 8.9), 0.22 mg/ml luminol (Sigma), 0.033 mg/ml paracoumaric acid, and 0.015% H₂O₂. The membrane was exposed to film for 1 or 5 min.

In vitro dimerization assay.

To transcribe RNA molecules consisting of the 5′ region of the HIV-1 vector with or without Ψ in vitro, a T7 promoter was attached by PCR to HIV sequences, starting from the R region of the 5′ LTR and ending 200 nucleotides downstream of the gag initiation codon. The PCR products were amplified, using either pHR′CMV-GFP or pHR′CMV-GFPΔΨ as a template and the T7HIV-TAR(F) and Gag9 primers (Table 1). The subsequent PCR products were transcribed in vitro, using an AmpliScribe T7-Flash transcription kit according to the manufacturer's instructions (Epicenter Technologies). The RNA was extracted with phenol-chloroform and was ethanol precipitated. An in vitro dimerization assay was carried out essentially as described before (43), with minor modifications. The RNA (25 μg/ml) was resuspended in 50 mM Na cacodylate (pH 7.5), denatured at 95°C for 10 s, and immediately cooled on ice. An equal volume of 2× dimer buffer (100 mM Na cacodylate, 10 mM MgCl₂, 600 mM KCl [pH 7.5]) was added to the samples (10 μl/sample), which were either kept on ice or incubated at 37°C for 3 h to allow dimer formation. The samples were electrophoresed on 1% agarose gels containing ethidium bromide.

Western blot analysis.

Protein analysis was performed using standard Western blot analysis as previously described (48). To detect capsid proteins, ascites-purified monoclonal anti-HIV-1 capsid antibody (hybridoma clone 183-H12-5C; NIH AIDS Research and Reference Program) was used at a 1:10,000 dilution. The secondary antibody was a horseradish peroxidase-conjugated polyclonal goat anti-mouse antibody (Jackson Immunoresearch Laboratories) used at a 1:10,000 dilution. Densitometric measurements of the capsid protein levels were performed using the TINA software program.

RESULTS

Construction of an HIV-1-based vector lacking the packaging signal.

During the course of our study of HIV-1-based vector transduction, we constructed an HIV-1 vector (pHR′) (52) lacking the core encapsidation signal. Specifically, a 105-bp-long deletion spanning the four stem-loops (SL1 to SL4) of the Ψ (Fig. 1A) was introduced into pHR′CMV-GFP (a pHR′ vector encoding wt EGFP) to create ΔΨ. Since the Ψ sequence includes the HIV-1 SD (located in SL2), deletion of this element may result in activation of cryptic splicing sites. To avoid this problem, we introduced the SD to the site of the deletion. A PmeI restriction site was also inserted to serve as a genetic marker for the presence of the ΔΨ mutation (Fig. 1B). Other cis-acting elements needed for vector expression and transduction were left unchanged.

Expression and transduction of the ΔΨ HIV vector.

To test the expression and transduction levels of the ΔΨ HIV vector, 293T cells were transfected with wt or ΔΨ plasmids in addition to helper plasmids. The latter plasmids included pCMVΔR8.2, which provides all the vector proteins except for the envelope and Vpu proteins, and pMD.G, which expresses the heterologous protein VSV-G, which pseudotypes the vector (51, 52). Equal amounts of supernatants from the transfected cultures were collected and incubated with naïve 293T cells, and 2 days postinfection, the infected cells were analyzed by FACS to detect GFP-positive cells. While the mutant vector was initially constructed to serve as a negative control in the transduction assays, we were surprised to observe significant transduction of this vector. Compared to that of the wt vector, transduction of the ΔΨ vector was less efficient; however, the reduction in infected cells was only twofold (Fig. 2A). To better quantify the transduction level of the ΔΨ vector, we repeated this experiment (n = 6) and examined the total GFP fluorescence intensity of the transfected cells (to monitor transfection efficiency), RT levels of the culture supernatants (to determine particle production), and total GFP fluorescence intensity of the infected cells (to monitor infection levels). Overall, the transfected cells showed fluorescence intensities of approximately one order higher than those of the infected cells, probably due to the episomal amplification of the transfected plasmid DNA, as the cells express the large T antigen and the plasmid backbones carry the simian virus 40 origin of replication. Importantly, the average expression levels of the transfected vectors were comparable. The percentage of GFP-positive cells for the wt was 93.2 ± 3.5% versus 90.7 ± 4.9% for ΔΨ, while the mean fluorescence intensity for the wt was 3657 ± 602 versus 2840 ± 657 for ΔΨ (the total GFP fluorescence intensity is presented in Fig. 2B). The average production level of the virions, monitored by exogenous RT assay (6.9 ± 1.6 arbitrary units for the wt versus 6.0 ± 1.7 for the ΔΨ vector media), was also comparable. When average infection levels were calculated, we consistently found that the infection level for the ΔΨ vector was only two- to threefold lower than that for the wt vector: the percentage of GFP-positive cells for the wt was 73.4 ± 3.9% versus 42.9 ± 6.2% for ΔΨ, while the mean fluorescence intensity for the wt was 499 ± 217 versus 325 ± 148 for ΔΨ. See Fig. 2B for the total GFP fluorescence intensities of infected cells. The calculated multiplicity of infection ranged between 1 to 2.4 for the wt, with a mean of 1.4 ± 0.2, and between 0.3 to 1.1 for ΔΨ, with a mean of 0.6 ± 0.1. Thus, although the core-packaging signal is deleted from ΔΨ, this vector seems to have significant levels of transduction.

FIG. 2.

Expression and transduction of an HIV-1-based vector with an encapsidation signal deletion. (A) 293T cells were transfected with helper plasmids pCMVΔR8.2 and pMD.G together with pHR′CMV-GFP (wt) or pHR′CMV-GFPΔΨ (ΔΨ) vector DNA. Two days posttransfection, culture media were collected and equal volumes (diluted 1:3 in the presence of 8 μg/ml polybrene) were used to infect naïve 293T cells. Infected cells (solid lines) were analyzed for GFP expression 2 days postinfection by FACS and were compared to mock-infected cells (dashed lines). (B) Graph summarizing total GFP expression (see Materials and Methods for calculation) in transfected (gray bars) or infected (black bars) 293T cells with either wt or ΔΨ vectors. The data are represented as the means ± the standard error of the means (n = 6) of the total GFP fluorescence intensities and are shown on a logarithmic scale with numerical annotation of the means.

We next analyzed whether the observed transduction of the ΔΨ-containing vector is limited to the GFP-containing vector. To test this, we introduced the ΔΨ mutation described above (Fig. 1) into an additional HIV-1-based vector, pHR′CMV-lacZ (52), to create pHR′CMV-lacZΔΨ. The lacZ vector is similar to the GFP vector but carries a significantly longer reporter gene (a 3-kb-long lacZ gene versus a 0.7-kb-long gfp gene). In a single-cycle infection assay similar to the one described above, we cotransfected pHR′CMV-lacZ and pHR′CMV-lacZΔΨ with the helper plasmids pCMVΔR8.2 and pMD.G and compared the transduction of these vectors, using a quantitative liquid β-galactosidase assay (66). The transduction levels, normalized to viral particle content as determined by exogenous RT activity, were calculated from four independent experiments and on average were found to be about fourfold lower for pHR′CMV-lacZΔΨ (7 ± 4 arbitrary units) than for pHR′CMV-lacZ (32 ± 9 arbitrary units). This result further demonstrates significant transduction of yet another vector with no core-packaging sequence.

Does transduction of the ΔΨ vector represent a true infectious event?

To unequivocally confirm that the observed fluorescence intensity of the cells infected with the ΔΨ vector was the result of genuine infection events, we tested various parameters. First, we checked whether the transduction of the ΔΨ vector was due to contamination of the culture supernatants from the carry-over of plasmid DNA (50), thus resulting in transfection rather than infection of the vector plasmids. We cotransfected the ΔΨ vector with the individual helper plasmids (pMD.G or pCMVΔR8.2) or with a mixture of both and incubated the subsequent culture media with naïve cells. Only media from cells cotransfected with the wt or ΔΨ vectors and the two helper plasmids could infect naïve 293T cells (Fig. 3A), indicating that GFP expression in infected cells is indeed dependent on the formation of infectious particles and not on the carry-over of GFP-expressing plasmid DNA.

FIG. 3.

Expression of the ΔΨ vector in infected cells represents genuine transduction. (A) FACS analysis of GFP expression following transfection (top panel) and infection (bottom panel) of 293T cells with the ΔΨ vector (solid black line). This vector was cotransfected with helper plasmids (pCMVΔR8.2 or pMD.G) as indicated. Cells were analyzed 2 days posttransfection, and their media (diluted 1:3 and supplemented with 8 μg/ml polybrene) were used to infect naïve 293T cells. Control cells (thin dashed line) were either mock-transfected or infected 293T cells. (B) FACS analysis of GFP expression in 293T cells infected with either wt or ΔΨ virions (solid lines) in the absence or presence of nevirapine (Nev; 12 μM). Cells were screened 5 days following infection. Dashed lines indicate control cells, as above. (C) Circle junction analysis of both the wt and ΔΨ vectors. Low-molecular-weight DNA was extracted, using the Hirt method, from 293T cells infected with media from cells expressing vector and pCMVΔR8.2 (1) or with vector and both pCMVΔR8.2 and pMD.G (2). The LTR-LTR junctions were amplified from the low-molecular-weight DNA samples, using PCR and primers derived from the U3 and R sequences in the LTRs (2LTRcircle-F and 2LTRcircle-R). The negative control (-) for PCR included all PCR components but no template DNA. (D) Maintenance of the wt packaging signal or the ΔΨ mutation in a transduced vector. The genomic DNA was extracted from the infected cells described for panel C. The 5′ portions of the vector sequences were detected by PCR, using primers (T7HIV5′R and Gag9) that flank the packaging signal or the deletion site. The identities of the amplified fragments were further confirmed by digestion with the indicated restriction enzymes. (E) Alu PCR analysis to detect integrated forms of wt and ΔΨ vectors. The genomic DNA extracted for panel D was used as a template in a nested PCR using primers derived from Alu and vector sequences. The amplified smear (left panel) was analyzed by Southern blotting (right panel) using a vector-specific probe. Cells infected with media lacking virions are shown in the lanes labeled “mock”.

A recent report (53) demonstrated conditions in which protein transduction rather than vector transduction can occur. To test whether this was the case in our system, we incubated cells with the media containing wt or ΔΨ particles in the absence or presence of the RT inhibitor nevirapine (12 μM) (Fig. 3B). In the absence of nevirapine, we observed a 2.3-fold reduction in transduction for the ΔΨ vector versus that for the wt, consistent with the above results. In the presence of nevirapine, transduction of both the wt and ΔΨ vectors was inhibited, suggesting that GFP fluorescence in infected cells is not the result of protein transduction but rather the outcome of vector transduction and GFP expression in these cells. In addition, we passaged wt- and ΔΨ-transduced cells for a period of 1 month; following nine passages in the absence of selection for the transduced cells, the ratio of total fluorescence intensity for ΔΨ versus that for wt was maintained (0.3 ± 0.012, n = 9), further indicating stable integration of the two kinds of vector.

Stable integration of the ΔΨ vector was further confirmed following transduction of cells with HIV wt- or ΔΨ-based vectors expressing the neomycin resistance gene. In a single-cycle infection assay similar to the one described above, we cotransfected plasmids pHR′CMV-IRES-neo and pHR′CMV-IRES-neoΔΨ with the helper plasmids pCMVΔR8.2 and pMD.G and compared the transduction levels of these vectors in HeLa cells. Following selection with Geneticin (G418), resistant clones were observed for both wt- and ΔΨ-transduced cells, confirming stable integration of the vectors.

We also tested for the presence of typical products of reverse transcription in the infected cells, namely, the presence of two-LTR circular viral DNAs. These structures are formed in the nuclei of infected cells by an intramolecule joining of the termini of a portion of the viral (or vector) linear DNA. We chose to analyze the presence of these circular products because they harbor a unique LTR-LTR junction that does not exist in the transfected plasmids, thus allowing their specific detection by PCR, even in the presence of contaminating plasmid DNA. To achieve this, we used the Hirt extraction method (35) to obtain low-molecular-weight DNA from infected cells and detected the LTR-LTR junctions in this preparation, using PCR with LTR-specific primers. This circle junction analysis amplified a PCR fragment with the expected size for both the wt and ΔΨ vectors (Fig. 3C). No products were observed in cells that were infected with media derived from transfections of the vector plasmids with pCMVΔR8.2 only. These data indicate that reverse transcription of the HIV-derived vectors, both wt and ΔΨ, occurred in the infected cells.

To further investigate the parameters of infection, we tested for the presence of vector DNA in the genomic DNA of infected cells and the maintenance of the Ψ deletion. High-molecular-weight genomic DNA was extracted from cells infected with wt and ΔΨ vectors, and the 5′ portion of the vector DNA was amplified by PCR. The primers used in the PCR spanned the Ψ sequences; hence, the PCR products had an expected size of 557 bp and 464 bp for the wt and ΔΨ vectors, respectively. As can be seen in Fig. 3D, PCR products with the expected sizes were observed for both the wt and ΔΨ vectors. To confirm the specificity of the PCR products, we digested them with BssHII or Pme1, since the BssHII recognition sequence exists only in the wt vector and the PmeI recognition sequence exists only in the ΔΨ vector (Fig. 1). Figure 3D demonstrates the differential digestion pattern of the PCR products, confirming the presence of the HIV-based vectors in the genomic DNA of infected cells and the maintenance of the Ψ deletion in the mutated vector. Further evidence for the presence of integrated forms of the vectors in the cellular genomic DNA was obtained following Alu PCR analysis of vector integration in infected cells (19). Fragments were amplified from genomic DNA of wt- and ΔΨ-infected cells, using a primer specific to the vector sequence and another specific to Alu sequences. These sequences were then analyzed for the presence of the specific vector by Southern blot hybridization, using a DIG-labeled RNA probe that includes the HIV-1 R and U5 sequences. Figure 3E represents both the ethidium bromide visualization of the nested Alu PCR (left panel) and the hybridized Southern membrane (right panel). Following nested Alu PCR, there was a large range of PCR fragments in all samples, including the mock-infected sample, which reflects amplification of genomic DNA fragments flanked by Alu sequences or amplification of Alu-LTR sequences. Southern hybridization with an HIV-1 LTR-based probe detected amplification of Alu-LTR sequences only in wt- and ΔΨ-infected cells. As expected for random integration events, amplification of Alu-LTR sequences produced a smear consistent with fragments of different sizes. This finding demonstrates that vector integration occurs for both wt and ΔΨ vectors in infected cells.

Overall, the above data support the observation that the ΔΨ HIV-based vector is packaged into viral particles and is transduced into naïve cells.

Is the ΔΨ HIV-derived vector packaged as a dimer?

The signal for dimerization is embedded within the Ψ signal (30, 56, 61); hence, deletion of the entire Ψ sequence also inherently deletes the DIS and is expected to prevent dimerization of the mutant RNA. To verify this, we performed an in vitro dimerization assay (43) whereby short fragments of the vector RNA that start at the R sequence and extend into the gag sequence, with or without the Ψ signal, were transcribed by T7 polymerase in vitro, heat denatured, and chilled to 0°C. The RNA molecules were then incubated in a dimerization buffer at 37°C to allow their dimerization or kept at 0°C to retain their monomeric status. Electrophoresis of the resulting RNA forms revealed that the Ψ-containing molecules were monomeric at 0°C but generated slower-migrating structures, resembling RNA dimers, after incubation at 37°C (Fig. 4A). We noted that the slower-migrating RNA were diffuse in nature and not as distinct as the ones described in the above publication (43). This variation may be accounted for by minor differences in the conditions of the assay (see Materials and Methods) and the different sequence compositions of the in vitro-transcribed RNA. Nevertheless, RNA molecules with the ΔΨ mutation appeared monomeric regardless of their incubation temperatures, in contrast to wt RNA. This result suggests that the Ψ deletion indeed prevents dimerization of the RNA molecules. We further studied the dimerization status of the vectors packaged in virions essentially as described before (28). We purified VLPs containing wt or ΔΨ vectors and extracted the vector RNA. Aliquots of the RNA were incubated at different temperatures and then detected by Northern blot analysis using nondenaturing gel and a vector-specific probe to compare the dimer-to-monomer dissociation (Fig. 4B). This analysis demonstrated that the wt vector exists as a dimer which begins to dissociate at 48°C, which is similar to past reports (28, 64). In contrast, the majority of ΔΨ-vector RNA does not exist as a normal, slower-migrating dimer form but appears as a faster-migrating species, even at subphysiological temperatures. These data indicate that the ΔΨ HIV-based vector is packaged as a monomer and not as a dimer.

FIG. 4.

Dimerization assays showing the absence of normal dimers in the ΔΨ vector. (A) In vitro dimerization assay. RNA fragments of wt and ΔΨ vectors encompassing sequences starting from the 5′ R and ending 200 nucleotides downstream of the gag initiation codon were transcribed in vitro. The RNA (25 μg/ml) was heat denatured at 95°C in dimerization buffer, then placed on ice immediately. A portion of the RNA was then incubated at 37°C to allow for dimer formation. RNA forms (250 ng/sample) were resolved in a 1% agarose gel containing ethidium bromide. (B) In vivo dimerization assay. Virions were purified from media of 293T cells cotransfected with either wt or ΔΨ and the helper plasmids through 25% sucrose cushions and ultracentrifugation. Virions were lysed with lysis buffer containing SDS, proteinase K, and tRNA. RNA was then purified with phenol-chloroform and ethanol precipitation. Equal-sized samples of wt and ΔΨ RNA were heated to the indicated temperatures for 10 min and separated on nondenaturing agarose gel. The RNA forms were detected by Northern blot analyses using a labeled antisense riboprobe (U3-R-gfp) harboring the vector-specific U3, R, and gfp sequences.

Is overproduction of the ΔΨ vector the driving force for its encapsidation?

The results so far indicated that HIV-1 ΔΨ vectors can be transduced efficiently, but the question of how such vector RNA gets packaged remained. One potential mechanism is that the overproduction of this RNA in 293T cells (due to their exceptionally high transfection rates and the subsequent episomal amplification of the vector DNA) results in nonspecific packaging due to the sheer abundance of vector RNA. Hence, we aimed to test the packaging of HIV vectors in cells expressing reduced levels of vector RNA. To do this, we tested the transduction of wt and ΔΨ vectors expressed from integrated rather than transfected forms. We previously observed that, in this condition, the expression level of the vector in infected cells is approximately one order of magnitude lower than that in transfected cells (Fig. 2B). Cells were infected with the wt or ΔΨ vector (primary infection), and 1 day after infection, these cells were transfected with the helper plasmids and the media were replaced. Two days posttransfection, the supernatants were harvested and the rescued vectors were utilized to infect naïve 293T cells (secondary infection). The fluorescence intensities of the GFP in the different cultures were determined by FACS, and the average fluorescence intensity for each condition was calculated. A comparison of the total fluorescence intensities of the cells following the secondary infection showed a reduction in transduction levels of the rescued ΔΨ vector (ranging between four- to sevenfold, n = 5), consistent with the reduction seen in the primary infection. These results indicate that there is a consistent ratio in the efficiency levels of packaging of wt versus ΔΨ vectors that is independent of the expression levels of the vector RNA.

In a second approach to test the effect of overexpression of the vector RNA on its packaging, we compared the packaging of HIV vectors to that of overexpressed, nonviral mRNA. We transfected 293T cells with plasmids expressing helper proteins together with wt, ΔΨ, or pEGFP-N3 DNA. The latter plasmid overexpresses mRNA with the same reporter gene (EGFP) but with no additional retroviral sequences. Two days posttransfection, VLPs were purified through sucrose cushions to assay the amount of the assembled Gag proteins. In addition, total RNA was extracted from the transfected cells and from VLPs, respectively, in the culture media to assay RNA expression and encapsidation. The purified RNA was treated with RNase-free DNaseI and reverse transcribed, and serial dilutions of the cDNAs were analyzed with GFP-specific primers by PCR. Figure 5 represents the data for one out of two independent experiments which demonstrated similar results. Semiquantitative analysis revealed that the RNA levels of the retroviral vectors (with or without the Ψ signal) were comparable in cells and were reduced approximately threefold for ΔΨ compared to that for the wt vector in virions (Fig. 5A). In contrast, the GFP mRNA (GFP-N3) derived from pEGFP-N3, which was expressed in the cells at higher levels than the retroviral vectors, was very poorly packaged, at levels approximately four orders of magnitude less than those for the wt HIV vector. This difference could not be accounted for by the reduced VLP production in the EGFP-N3 sample that we observed in the Western blot of capsid protein (Fig. 5B), since the amount of VLP was approximately only twofold less than for HIV vector samples. To quantify the GFP RNA amount more accurately, the cellular and VLP RNA levels were analyzed by qPCR for each transfection (Fig. 5A). The ratio of virion to cellular RNA for each sample was then compared to that for the wt sample. Furthermore, to compensate for the minor variation in VLP production (Fig. 5B), the RNA ratio was normalized to VLP levels. The results of this analysis revealed that the difference between packaging efficiencies of the wt and ΔΨ vectors was only 5.5-fold, whereas the GFP mRNA, derived from pEGFP-N3, was packaged at a level 2.1 × 10⁴-fold less than that for the wt vector. In this system, we also tested encapsidation of a MoMLV-based vector harboring the GFP reporter gene (Mvec) in HIV-1 particles. The results of this analysis showed that Mvec was less efficiently packaged than both HIV-1 vectors, with a 16.5-fold reduction compared to the wt and a 3-fold reduction compared to the ΔΨ vector (Fig. 5A and B). However, Mvec was packaged with an approximately two orders of magnitude higher level of efficiency than that of pEGFP-N3-derived mRNA. As expected, due to the incompatibility of the Mvec and HIV proteins, this vector, although packaged, failed to transduce naïve 293T cells (data not shown).

FIG. 5.

Specificity of vector RNA packaging. (A) Semiquantitative PCR was performed to detect packaged RNA in VLPs from producer cells that were transfected with helper plasmids and either pHR′CMV-GFP (wt), pHR′CMV-GFPΔΨ (ΔΨ), pEGFP-N3 (pGFP) or pQCXIP-EGFP (Mvec). The lanes labeled “mock” show control cells transfected with no DNA. RNA was extracted from the producer cells and VLP-containing media, treated with DNase, and then reverse transcribed. The cDNA (marked as RT-PCR) was serially diluted at a ratio of 1:10, as were equal-sized samples of negative controls consisting of matched RNA (marked as PCR). All these samples were then PCR amplified using primers specific to GFP. Real-time qPCR was performed on the same samples, and the values shown are standardized to the amount of wt RNA for cells and virions. (B) Western blot analysis of HIV capsid protein. Virions from the samples transfected for panel A were purified through 25% sucrose cushions and ultracentrifugation. Virion pellets were analyzed by Western blot analysis with anti-CA antibodies. The intensities of the bands were quantified by densitometry, and the values were standardized to that for the wt (shown below the panel). (C) Semiquantitative and quantitative PCRs were performed specifically to detect viral genomic RNA in the samples described for panel A. ND, not detected. The star indicates an empty lane with a spill-over from the adjacent lane.

The above analyses were based on amplification of a GFP-derived sequence, because this marker is shared by all the tested retroviral vectors and by the mammalian expression construct pEGFP-N3. However, this analysis detects both kinds of RNA generated in the cells, subgenomic mRNA (controlled by an internal cytomegalovirus [CMV] promoter) (Fig. 1A) and genomic vector RNA. To differentiate between these two RNA species and to specifically study genomic RNA packaging, we subjected the retroviral vector RNA to RT-PCR analysis using genome-specific primers able to amplify common sequences in both the MoMLV and HIV vectors. One of these primers is positioned in the internal CMV promoter sequence upstream of the transcription initiation site of the subgenomic RNA, and the second primer is derived from the gfp gene. Semiquantitative and qPCR analyses with these primers demonstrated that the genomic RNA of the ΔΨ and MoMLV vectors, respectively, were packaged 2.4- and 12-fold less than the wt vector (Fig. 5C), consistent with the data listed above. Equivalent results were seen in an additional independent experiment in which cellular expression of the vector RNA was equivalent for both HIV vectors and the MoMLV vector (data not shown).

Altogether, these results demonstrate that, in HIV-1 particles, the packaging of an HIV vector with its packaging and dimerization signals deleted is more efficient than the packaging of a MoMLV-based vector. Yet even the latter retroviral RNA can be packaged at higher levels than a cellular overexpressed mRNA (EGFP). These results clearly indicate that overexpression per se is not sufficient to achieve efficient packaging in this system. Overall, these data indicate that the packaging of ΔΨ vectors is influenced by factors other than its levels of expression.

DISCUSSION

This study reveals that the transduction of HIV-1 vectors with their packaging and dimerization signals deleted is feasible and can occur with reasonable efficiency. We deleted the SL1 to SL4 core Ψ sequence, including the DIS signal, from vectors that carry the lacZ, neo, or gfp reporter genes and analyzed the transduction of these mutant vectors in single-cycle infection assays. We observed that the ΔΨ mutant vectors could be transduced with considerable efficiency, showing only a two- to fivefold reduction in efficiency compared to the wt vectors.

The majority of these analyses relied on the quantification of GFP fluorescence in cells infected by the vector. While this is a common, highly sensitive method, it has been suggested that it can reflect the transfer of GFP molecules associated with the viral particles rather than transduction of the vector itself (53). However, several lines of evidence indicate that the mutated vectors truly transduce the infected cells. First, whereas the transfer of GFP molecules is marked by a decrease in the fluorescence of the infected cells over time (53), the fluorescence of the cultures infected by either the wt or ΔΨ mutant vector actually increased over time (data not shown), indicating de novo synthesis of the GFP from the retroviral vector. In addition, transduced cells were maintained in culture for 1 month, during which time the ratio of the GFP expression level of the wt to that of ΔΨ was constant. Second, the presence of the RT inhibitor nevirapine eliminated GFP fluorescence in cultures infected by both the wt and ΔΨ mutant vectors. This finding suggests that this fluorescence originates from GFP molecules that are expressed from products of reverse transcription and not from the carry-over of preexisting proteins or plasmid DNAs from producer cells. Third, cells were transduced with HIV-based vectors expressing the neomycin resistance gene, and stable Geneticin-resistant clones were observed for both the wt and ΔΨ vectors, suggesting stable integration of the vectors. Fourth, typical reverse transcription products of both the wt and mutant vectors could be identified by PCR in the infected cells; LTR-LTR junctions and internal vector sequences, respectively, could readily be amplified from low- and high-molecular-weight DNA preparations. Fifth, random integration sites of the wt and ΔΨ mutant vectors could be detected in the genomes of the infected cells by Alu PCR and Southern blot analysis. Altogether, these results demonstrate that HIV vectors with their packaging and dimerization signals deleted can complete a full, single cycle of infection.

Many previous studies have demonstrated the importance of the Ψ signal for efficient packaging and replication of the HIV-1 genomic RNA (reviewed in references 11, 24, 56, and 61), thus establishing the idea that this sequence is a crucial cis-acting signal that should be included in vectors that are derived from this retrovirus. These studies included mutagenesis of different parts of the four stem-loops that consist of the core-packaging signal, many of which caused significant reductions in the packaging levels of the genomic RNA of HIV-1, particularly if the mutations spanned SL1 to SL3 (1, 18, 32, 40, 41, 45, 46, 59, 62). The documented effect on packaging varied and could reach up to a 100-fold reduction (1) in packaging of mutant versus wt genomes. In several studies that also investigated the effect of Ψ mutations on transduction, parallel reductions in infectivity were observed with reduced encapsidation (32, 59), and in some cases, the Ψ mutants were unable to replicate in spreading assays (1, 40). In light of these results, the fact that adequate infections could be observed for vectors generated in this study with all four of the stem-loops of Ψ deleted can be viewed as surprising. However, there is a major difference between previous publications and our current study in that all the above studies based their deletions on full or partially altered genomes expressed from a single plasmid, whereas our deletions were based on the vector system, which involves separation of cis- and trans-acting elements to different plasmids. The latter system avoided the pleiotropic effects such as reduced Gag expression or processing that were previously reported for some of the packaging signal mutations (32, 40, 62). Furthermore, note that although a large range of effects on encapsidation was reported for the ΔΨ mutants, in the cases where normalized amounts of virions were studied the difference in packaging was only two- to fivefold (41, 45, 59, 62), which is in strong agreement with our results. Importantly, in our study the reduction in transduction approximately correlated with the observed reduction in packaging, suggesting that no other steps in the cycle of infection were affected by the ΔΨ mutation.

Deletion of DIS from the ΔΨ vector abrogated the normal dimerization of the vector RNA, as indicated by the failure of ΔΨ to form proper dimers in vitro and in vivo. Although we cannot exclude the possibility that the mutant RNA formed weak dimers that had dissociated during the purification steps, we did not observe any slow-migrating dimer forms for ΔΨ RNA, while slower-migrating dimer forms were seen for wt RNA. In addition, while the wt RNA dimers could be dissociated to faster-migrating monomer forms at temperatures higher than 44°C, ΔΨ RNA had the same forms at all the examined temperatures, with no apparent dissociation with increasing temperatures. Note that a significant portion of the ΔΨ RNA showed intermediate levels of migration and failed to dissociate even at 60°C. It is not clear what caused this aberrant migration. It is possible that ΔΨ RNA forms some abnormal inter- and/or intramolecular connection; the R/U5 region may play a role in such interactions of the ΔΨ vector, since it was suggested that it forms an additional contact point for dimerization (36). In principle, the palindromic PmeI restriction site that replaced the natural DIS in the ΔΨ vectors may also contribute to such interactions. The AT-rich sequence (GTTTAAAC) of the PmeI restriction site replaced a GCGCGC palindromic loop in the context of SL1; however, this replacement lacks the natural stem-loop context of SL1. Hence, it is unlikely to contribute significantly to dimerization. Ongoing studies are investigating the effect of PmeI on dimerization; however, preliminary results indicate that a modification of the PmeI site by insertion of a linker (TTCCTTCCTTCC) has only a twofold effect on ΔΨ transduction (data not shown). Other sequences downstream of SL1/DIS (nucleotides 311 to 415) have been reported to contribute to dimerization in vitro in the presence (20) or absence (43) of the NC protein; however, these are also deleted in our ΔΨ vectors.

Although we failed to see normal dimerization in the ΔΨ mutants, they were infectious. Previously, Sakuragi et al. (64) noted more than a 100-fold reduction in infection of particles harboring monomeric forms of the vector RNA; however, as was reported by the authors, this reduction probably reflected the specific mutation used in their study (duplication of a 1.1-kb fragment containing an additional packaging signal) rather than the lack of dimerization itself. Similarly, Shen et al. (68) reported that a mutation in SL1 partially reduced genome dimerization and a late step of reverse transcription, yet as suggested by the authors, this specific mutation may have affected the general structure of the primer binding site region, resulting in aberrant strand transfer. In our study, the reduced level of transduction of the ΔΨ vectors appeared to be attributed to the equally reduced level of packaging, while the failure to form normal dimers did not have an additional inhibitory effect on transduction, suggesting that normal dimerization is not a prerequisite of reverse transcription.

The question of what drives the packaging of a ΔΨ HIV-1-based vector in our system remains. One important consideration is the effect of the overexpression of the vector and its helper proteins in the highly expressing 293T cells. Perhaps the sheer abundance of vector RNA in 293T cells leads to the encapsidation of the vector RNA into the virions. If this random packaging is in effect, then any equally abundant RNA should also be packaged. To test this, we compared the packaging levels of wt and ΔΨ vector RNAs containing the GFP reporter gene to the level of GFP mRNA expressed from a standard mammalian expression plasmid. Whereas ΔΨ RNA was packaged fivefold less than wt, as determined by semiquantitative RT-PCR and real-time qPCR, abundant GFP mRNA was packaged at a level four orders less than that for the wt, clearly demonstrating that the presence of overabundant RNA is not the mechanism of packaging of the ΔΨ vector. This finding was further supported by rescue experiments whereby we tested the packaging of wt and ΔΨ vectors from cells infected, rather than transfected, with these vectors. The infected cells expressed tenfold-less RNA than the transfected cells, providing an alternative system to study the effect of RNA abundance on packaging. In this system, we observed a four- to sevenfold reduction in the rescue of ΔΨ versus that for the wt, consistent with the difference observed in conditions where higher levels of expression of the vector is achieved (transfected cells). Overall, we concluded that overexpression of the mutated vector is not the driving force for its encapsidation.

The mechanism of ΔΨ packaging may depend on HIV vector sequences other than SL1 to SL4. Previous studies (17, 22, 34, 47, 60, 65) have reported significant reductions in dimerization, encapsidation, and/or infection levels following deletions in TAR, R/U5, or U5/L, suggesting a significant contribution of these elements to packaging. In addition, one study reported the contribution of the SD specifically to encapsidation: whereas an SL1-to-SL3 deletion that included the SD site resulted in a 50-fold reduction in encapsidation levels, adding back a heterologous SD site resulted in only a 12-fold reduction in encapsidation levels (46). Thus, the SD site that was retained in our ΔΨ mutants could contribute to the minimal decrease in encapsidation and transduction of these vectors. Furthermore, other observations demonstrated that sequences downstream of the gag initiation codon improve packaging and/or transduction (15, 41, 58). Although the gag ATG is deleted in our construct, partial downstream sequences are present (51, 52); hence, packaging may be facilitated by these sequences. Indeed, in preliminary experiments further deletion of all of the gag sequences as well as the SD site abolished transduction (data not shown). Therefore, a combination of intact sequences upstream and downstream of our Ψ deletion and inclusion of the SD site may contribute to packaging of ΔΨ vector RNA.

Packaging of the retroviral vector RNA by HIV-1 Gag appeared to extend to packaging of MoMLV. This was deduced from experiments that paralleled those above, in which we investigated the packaging of a MoMLV-based vector, which also encodes the GFP reporter gene. In this instance, we observed a 17-fold reduction in the level of encapsidation compared to that for wt HIV-based vector RNA; however, this equates to approximately a 300-fold improvement in the level of packaging over that for GFP-N3 mRNA. Packaging of MoMLV-based vector RNA is probably facilitated by the ability of HIV Gag to bind MoMLV RNA, which has previously been demonstrated in vitro (13) and in vivo (42). The above explanation relies on specific cis-acting signals that directly contribute to packaging. An alternative and not mutually exclusive explanation is that cis-acting signals govern the trafficking and compartmentalization of the vector RNA. If this is the case, the HIV and MoMLV genomic RNAs, but not the cellular GFP-N3 mRNA, may be targeted to the same specific intracellular pathway and/or compartment(s) for packaging by Gag proteins (recently reviewed in reference 71), hence facilitating their encapsidation. For example, the Rev/RRE system in HIV is a good candidate for dictating the fate of the genomic RNA in the cell; indeed, this system has recently been shown to influence not only the nuclear export of the genomic RNA but to also directly enhance its packaging (14).

The ΔΨ mutation may contribute to the safe use of HIV-1 based vectors. It has been reported that DIS serves as a hotspot for recombination (7); hence, removal of this signal in the ΔΨ vectors should reduce recombination. More importantly, Ψ-gag recombination can routinely occur during the generation of HIV-1 based transducing vectors (67). Full elimination of the Ψ signal in the system reported here prevents such recombinations. Thus, in situations where suboptimal transductions are acceptable, the ΔΨ mutation can be used with enhanced safety. Overall, we demonstrated that transduction of HIV-1-based vectors lacking the Ψ and DIS signals is feasible. Furthermore, the absence of these cis-acting signals enabled us to formally demonstrate that the Ψ and DIS sequences do not contribute to the early stages of vector transduction.