Construction and In Vitro Properties of a Series of Attenuated Simian Immunodeficiency Viruses with All Accessory Genes Deleted

Abstract

We have generated simplified simian immunodeficiency virus (SIV) constructs lacking the nef, vpr, vpx, vif, tat, and rev genes (Δ6 viruses). To accomplish this, we began with an infectious molecular clone of SIV, i.e. SIVmac239, and replaced the deleted segments with three alternate elements: (i) a constitutive transport element (CTE) derived from simian retrovirus type 1 to replace the Rev/Rev-responsive element (RRE) posttranscriptional regulation system, (ii) a chimeric SIV long terminal repeat (LTR) containing a cytomegalovirus (CMV) promoter to augment transcription and virus production, and (iii) an internal ribosome entry site (IRES) upstream of the env gene to ensure expression of envelope proteins. This simplified construct (Δ6CCI) efficiently produced all viral structural proteins, and mature virions possessed morphology typical of wild-type virus. It was also observed that deletion of the six accessory genes dramatically affected both the specificity and efficiency of packaging of SIV genomic RNA into virions. However, the presence of both the CTE and the chimeric CMV promoter increased the specificity of viral genomic RNA packaging, while the presence of the IRES augmented packaging efficiency. The Δ6CCI virus was extremely attenuated in replication capacity yet retained infectiousness for CEMx174 and MT4 cells. We also generated constructs that retained either the rev gene or both the rev and vif genes and showed that these viruses, when complemented by the CMV promoter, i.e., Δ5-CMV and Δ4-CMV, were able to replicate in MT4 cells with moderate and high-level efficiency, respectively. Long-term culture of each of these constructs over 6 months revealed no potential for reversion. We hope to shortly evaluate these simplified constructs in rhesus macaques to determine their long-term safety as well as ability to induce protective immune responsiveness as proviral DNA vaccines.

A major advantage of an attenuated virus strategy for the development of a human immunodeficiency virus (HIV) vaccine might be the ability of live attenuated viruses to induce broad and persistent immunity. The existence of long-term nonprogressors in regard to HIV type 1 (HIV-1) infection (13, 26, 32, 35, 42, 43, 51) and of multiply exposed, uninfected individuals (10, 15, 16, 24, 33, 34, 39, 40) suggests that naturally attenuated species might exist and play a protective role for at least a transient period. Similarly, inoculation of macaques with attenuated variants of simian immunodeficiency virus (SIV), containing deletions in nonessential genes, has yielded protection against subsequent challenge by virulent SIV strains (1, 12, 14, 22, 23, 50).

However, important safety concerns have limited the application of these findings in HIV vaccine research, because even multiply deleted SIV constructs, containing deletions of nef, vpr, and the negative regulatory element (NRE), were pathogenic in both infant and adult macaques (2, 21). Long-term human nonprogressors known to be infected by nef deletion variants of HIV were shown to have falling CD4 counts and rising viral loads, accompanied by disease progression, over time (19, 28). Therefore, live attenuated primate lentiviruses that contain deletions of nonessential genes may harbor residual potential for pathogenesis.

The SIV macaque model has proven invaluable (6, 36), yet basic research on SIV has been limited in comparison with that performed with HIV. It is notable that all of the above-mentioned live attenuated SIVs except those mutated in vif have retained efficient replication capacity in permissive cell lines (18). Indeed, even SIVs defective in the NRE, nef, vpr, vpx, and vif (Δ5 viruses) were able to replicate to high titers after long-term passage in CEMx174 cells (18). This finding may account for the fact that even the highly attenuated Δ3 virus, lacking nef, vpr, and the NRE, can cause disease, since quickly replicating viruses probably retain considerable capacity for compensatory mutagenesis. In addition, all of these mutated viruses retained the two important regulatory genes, i.e., tat and rev, known to be essential for the efficient replication of both HIV and SIV. Indeed, the presence of tat is strongly linked to viral pathogenesis, and strong immune responses to both Tat and Rev are correlated with nonprogression (47, 52). Both Tat and Rev may also have adverse effects on the host (17, 30, 38). Therefore, a safe, live attenuated vaccine might even require that both tat and rev be deleted.

At the same time, research has shown that defective tat viruses can be partially corrected by the replacement of regulatory elements within the upstream part of the long terminal repeat (LTR) by a stronger cytomegalovirus (CMV) promoter (8). Replication-competent rev-negative viruses have also been independently generated, using a constitutive transport element (CTE) derived from simian retrovirus type 1 (SRV-1) to replace the Rev/Rev-responsive element (RRE) system (48, 53). Therefore, it should be theoretically possible to construct a simplified SIV that is devoid of all accessory genes; the question that remains is whether such viruses will retain pathogenic potential.

Here we describe the construction and characterization of a series of such simplified SIVs, using the molecular SIVmac239 clone as an initial genome. We have generated a Δ6 virus lacking nef, vpr, vpx, vif, tat, and rev through a series of large deletions. Select functional elements, such as a CTE, a chimeric SIV LTR containing a CMV promoter, and an internal ribosome entry site (IRES), have been introduced into our simplified SIV vectors to increase the production of viral structural proteins in the absence of accessory genes. These simplified SIV constructs can also form mature virions that possess wild-type morphology after transfection into COS-7 cells. Although the deletion of all viral accessory genes affected both the specificity and efficiency of viral genomic RNA packaging into virions, this defect was repaired by insertion of the CTE, the chimeric CMV promoter, and the IRES into the viral genome. With the help of the CMV promoter, a variant that retained the rev gene, Δ5-CMV, was able to persistently replicate in MT4 cells, while a variant that retained both the rev and vif genes, Δ4-CMV, was able to efficiently replicate in this cell line. Most importantly, these simplified SIVs were extremely attenuated in replication capacity yet remained infectious in CEMx174 cells, with no evidence of reversion after 6 months in tissue culture. We next wish to evaluate the safety and protective efficacy of these constructs in rhesus macaques while at the same time studying the in vivo replication capacity of these viruses to gain further insights into the specific roles of SIV accessory proteins as determinants of pathogenesis.

(Research performed by James B. Whitney for this study was in partial fulfillment of the requirements for a Ph.D. degree, Faculty of Graduate Studies and Research, McGill University, Montreal, Quebec, Canada.)

MATERIALS AND METHODS

Generation of SIV constructs.

The full-length infectious wild-type clone of SIV, SIVmac239/WT (20, 25), was used to generate all constructs described in this report (Fig. 1). Both a PCR-based mutagenesis method and Pfu polymerase were used to generate the deletions shown. In the case of the SIVmac239Δ5 mutant, two distinct regions were deleted. First, the sequences between the SphI and the BglII sites (positions 6702 to 4952) were replaced with the PCR fragments amplified by primers Svif (5′-GGCGCATGCGTCGACTCTGCTACCTCTCTAGCCTCTCCG-3′) and Sint1 (5′-CCCAGAATAGTGGCCTGATAGATAGTAGACACCTGTG-3), resulting in deletion of vif, vpx, vpr, and tat. Second, the region between the SacI and XhoI sites (positions 9482 to 10535) was replaced with the PCR fragments amplified by primers Senf-1 (5′-GGCGAGCTCACTCTCTTGTGATTGGCAATAGACATGTCTC-3′) and SU5-1 (20) to delete the nef gene. The resulting construct, SIVmac239Δ5, was then used to generate the SIVmac239Δ6 mutant by replacing the region between the SphI and HindIII sites (positions 6702 to 7079) with the PCR fragments amplified by primers Senv-1 (5′-GGCGGCATGCATGGGGTGTCTTGGTAATCAGCTGCTTATCGCC-3′) and Sen1 (5′-GCCATACATCCTCTATTGCCTG-3′) to delete both the tat and rev genes. The SIVmac239Δ4 mutant, lacking tat, nef, vpr, and vpx but not rev or vif, was generated similarly to SIVmac239Δ5 except that the region between the SphI and SacI sites (positions 6702 to 6011) was replaced with PCR fragments amplified by primers Svif-2 (5′-GGCGCATGCATCATGCCAGTATTCCC-3′) and Svif-4 (5′-CAAAGATTATGGAGGAGGAAAAGAGGTGG-3).

FIG. 1.

Schematic illustration of the SIV constructs generated. All enzyme sites used are indicated, and both deletions () and alternative elements () are shown. (A) Construction of simplified SIVs. All mutants contain two deletions. In the case of the SIVmac239 Δ4 construct, one deletion involves the vpx, vpr, and tat genes (from positions 6241 to 6702), while the other results in inactivation of nef (positions 9500 to 9674). The Δ5 and Δ6 constructs are identical to Δ4 except that the first deletion was extended to also delete the vif gene (positions 5667 to 6702) and both the vif and rev genes (positions 5667 to 6859), respectively. To generate Δ6-CTE, a 173-bp CTE of SRV-1 was inserted into the Δ6 vector at the position of the nef deletion. The Δ6-CTE-CMV construct was derived from Δ6-CTE by replacing both the 5′ and 3′ LTRs with a chimeric LTR containing the CMV IE promoter. The Δ6-CTE-CMV-IRES (Δ6CCI) construct was generated by insertion of an IRES element immediately upstream of the env gene in the Δ6-CTE-CMV vector. The Δ6CCI constructs that contained a poliovirus-derived IRES and an ECMV-derived IRES are designated Δ6CCI-P and Δ6CCI-E, respectively. The Δ5CCI construct is identical to Δ6CCI-P except that the vif gene is retained. WT, wild type. (B) Construction of SIV mutants that retain the rev gene. SIVmac239Δnef-CMV contains both the chimeric CMV-LTR insert and the nef deletion, while the Δ2-CMV construct contains additional mutations in the first exon of tat (which truncates the tat gene). Δ4-CMV and Δ5- CMV are identical to Δ4 and Δ5 except that both the 5′ and 3′ LTRs were replaced with the chimeric CMV-LTR. (C) Construction of the simplified SIV vector, SIVmac239Δ6-CTE-CMV-IRES-GFP (Δ6CCI-GFP), containing the EGFP reporter gene. Similar to the case of Δ6CCI, the Δ6CCI-GFP constructs that contained a poliovirus-derived IRES and an ECMV-derived IRES are termed Δ6CCI-P-GFP and Δ6CCI-E-GFP, respectively.

A CTE was inserted into the region of the deleted nef gene in the SIVmac239Δ6 construct to generate SIVmac239Δ6-CTE. Toward this end, the CTE was amplified from plasmid p72S240 (41), using primers CTE-1 (5′-GGCGAGCTCACTCTCTTGTGAATAGACCACCTCCCCTGCG-3′) and CTE-2 (5′-GAGACATGTCTATTGCCAACAAATCCCTCGGAAGCTGCG-3′). The PCR products were then used as megaprimers paired with primer SU5-1. The resulting PCR fragments were used to replace the region between the SacI and XhoI sites in SIVmac239Δ6.

For construction of the chimeric LTR construct termed CMV-LTR, the CMV promoter was first amplified from the vector termed pIRES-EGFP (Clontech), using primers cmv-1 (5′-GGAAGGGATTTATTACAGTGCCGCGTTACATAACTTACGG-3′) and cmv-2 (5′-GAATACAGAGCGAAATGCAGTGCTTATATAGACCTCCCACCG-3′). The resulted CMV fragments were then used as megaprimers paired with primer SU5-1 or CTE-1 to generate fragments CMV-U5-1 and CTE-CMV, respectively, based on Δ6-CTE. The two fragments were combined and used as the template to amplify the final fragment, CTE-CMV-LTR, using primers CTE-1 and SU5-1. The CTE-CMV-LTR fragment was then used to replace the region between the SacI and XhoI sites (positions 9482 to 10535) in SIVmac239Δ6 to generate the intermediate plasmid SIVmac239Δ6-CTE-3′CMV. A 5′-CMV-LTR fragment was generated by PCR based on CTE-CMV-LTR, using primers pSU3 and SU5-2 (5′-GTTCAGGCGCCAATCTGCTAGGGATTTTCCTGCTTCGG-3′). This fragment was then cloned into the EcoRI-NarI site of the SIVmac239Δ6-CTE-3′CMV vector to generate the SIVmac239Δ6-CTE-CMV (Δ6CC) construct.

The SIVmac239Δ6-CTE-CMV-IRES (Δ6CCI) mutant was generated based on the Δ6CC construct. First, an NcoI-HindIII fragment was generated by PCR using primers Senv-Nco (5′-GTGCCATGGGGTGTCTTGGTAATCAGCTGCTTATCGCC-3′) and Sen1. The IRES of encephalomyocarditis virus (ECMV) was cut out from the pIRES-EGFP vector (Clontech) using enzymes SphI and NcoI; then the ligation product of these two fragments was inserted into the SphI-HindIII site (positions 6702 to 7079) of the Δ6CC vector to generate the Δ6CCI-E construct containing the IRES of EMCV. Another construct containing the IRES of poliovirus, Δ6CCI-P, was generated using this same strategy except that the SphI-NcoI fragment of the IRES was produced by PCR from plasmid pCDNA3-rLuc-polIRES-fLuc using primers polio-1 (5′-GCAGCATGCTCTGGGGTTGTTCCCACC-3′) and polio-2 (5′-GCACCATGGCCGGATGGCCAATCCAA-3′).

To construct the SIVmac239Δnef-CMV mutant (Fig. 1B), the 5′ CMV-LTR of Δ6CC was used to replace the EcoRI-NarI region of the wild-type vector. Then the 3′ LTR (SacI to XhoI, positions 9482 to 10535 in this wild-type vector) was replaced with a chimeric 3′ CMV-LTR fragment generated in the same way as the CTE-CMV-LTR fragment described above except that primer Senf-1 was used instead of CTE-1 and vector Δ6 was used as template for PCR instead of Δ6-CTE. To generate Δ5-CMV and Δ4-CMV, both the 5′ and 3′ LTRs were similarly replaced with the chimeric CMV-LTR. The Δ2-CMV construct was made by replacing the region between the SphI and HindIII sites (positions 6702 to 7079) of SIVmac239Δnef-CMV with the PCR fragments amplified by primers Srev-1 (5′-GAAGCATGCTATAACTGATGATATTGTAAAAAGTGTTGC-3′) and Sen1 (5′-GCCATACATCCTC TATTGCCTG-3′) to introduce double stop codons in the first exon of tat without affecting the overlapping vpr and rev genes.

The SIVmac239Δ6-CTE-CMV-IRES-GFP (Δ6CCI-GFP) mutant (Fig. 1C), containing the enhanced green fluorescence protein (EGFP) reporter gene downstream of the IRES, was generated based on the Δ6CC construct. The fragment produced by SphI-SacI in the envelope gene in the Δ6CC construct was replaced with the SphI-SacI fragment containing the IRES and the EGFP region from the pIRES-EGFP vector (Clontech) to generate the Δ6CCI-E-GFP construct (in which E designates ECMV), containing the IRES of ECMV. Alternatively, the ligation product of the poliovirus IRES (SphI-NcoI) and the EGFP gene (NcoI-SacI) was used to replace the SphI-SacI region in the Δ6CC vector to yield the Δ6CCI-P-GFP construct (in which P designates poliovirus), containing the IRES of poliovirus.

All constructs were sequenced to confirm the validity of all fragments derived by PCR. Nucleotide designations are based on published sequences (25).

Cells and preparation of virus stocks.

COS-7 cells were maintained in Dulbecco modified Eagle medium (DMEM) supplemented with 10% heat-inactivated fetal bovine serum. CEMx174 and MT4 cells were maintained in RPMI 1640 medium supplemented with 10% heat-inactivated fetal bovine serum. Molecular constructs were purified using a maxi plasmid kit (Qiagen Inc., Mississauga, Ontario, Canada). COS-7 cells were transfected using these constructs with Lipofectamine-Plus reagent (GIBCO, Burlington, Ontario, Canada). Virus-containing culture fluids were harvested at 60 h after transfection and were clarified by centrifugation for 30 min at 4°C at 3,000 rpm in a Beckman GS-6R centrifuge. Viral stocks were passed through a 0.2-μm-pore-size filter and stored in 0.5- or 1-ml aliquots at −70°C. Levels of viral reverse transcriptase (RT) were determined as described previously (29), and levels of viral capsid antigen, i.e., p27, were quantified by a Coulter SIV core antigen assay kit (Immunotech Inc. Westbrook, Maine).

Viral protein analysis by radiolabeling and immunoprecipitation.

COS-7 cells were transfected with wild-type or mutant constructs. At 20 h after transfection, cells were starved at 37°C for 30 min in DMEM without Met and Cys. Radiolabeling was performed with [³⁵S]Met and [³⁵S]Cys at a concentration of 100 μCi/ml for 30 min at 37°C. Then the cells were thoroughly washed with complete DMEM and cultured for 1 h. Culture fluids were collected and clarified on a Beckman GS-6R bench centrifuge at 3,000 rpm for 30 min at 4°C. Viral particles were further purified through a 20% sucrose cushion at 40,000 rpm for 1 h at 4°C, using a SW41 rotor in a Beckman L8-M ultracentrifuge. Virus pellets were suspended in 1× sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) loading buffer (31), boiled for 5 min, then fractionated by electrophoresis on an SDS–12% polyacrylamide gel, and exposed to X-ray film. The labeled cells were washed twice with cold phosphate-buffered saline (PBS) and lysed in buffer containing 0.1% NP-40. Cell lysates were incubated with a monoclonal antibody (MAb) against SIV p27 at 4°C for 30 min, and the resultant antigen-antibody complexes were precipitated by a 30-min incubation with protein A-Sepharose CL-4B (Amersham-Pharmacia Biotech, Montreal, Quebec, Canada). The recovered viral proteins were analyzed by SDS-PAGE (12% gel) and exposed to X-ray film (31).

Virion morphology.

The morphology of the viruses produced by the various constructs described above was examined by transmission electron microscopy. Briefly, COS-7 cells transfected with wild-type constructs or the simplified SIV constructs were fixed after 40 h with 2.5% glutaraldehyde followed by 4% osmium tetroxide. Thin-sectioned samples were stained with lead citrate and uranyl acetate and visualized using a JEOL 200 FX electron microscope as described elsewhere (44).

Packaging of viral genomic RNA.

Viral RNA was isolated from equivalent amounts of COS-7 cell-derived viral preparations, based on levels of SIV p27 antigen, using a QIAamp viral RNA mini kit (Qiagen). RNA samples were treated with RNase-free DNase I at 37°C for 30 min to eliminate possible DNA contamination. DNase I was then inactivated by incubation at 75°C for 10 min. The viral RNA samples were quantified by RT-PCR, using the Titan One Tube RT-PCR system (Boehringer-Mannheim, Montreal, Quebec, Canada) as described previously (20) except that two pairs of primers were used in tandem. The primer pairs sg1 (5′GAAGCATGTAGTATGGGCAG-3′) and sg2 (5′GGCACTAATGGAGCTAAGACCG-3′) were used to amplify a 114-bp fragment representing the full-length viral genome. Another pair of primers, Senf-3 (5′-GGAAGATGGATACTCGCAATCC-3′) and SU3-3 (5′-GCACTGTAATAAATCCCTTCCAG-3′), was used to amplify a fragment between the end of the env gene and the beginning of the 3′ U3 region, which represents total viral RNA. The size of the Senf-3–SU3-3 product is 317 bp in the case of the wild-type viral genome, 142 bp in the case of the Δ6 genome, and 315 bp for each of the other constructs. Relative amounts of product were quantified by molecular imaging (Bio-Rad, Toronto, Ontario, Canada).

Virus infection.

Viral stocks were thawed and treated with 100 U of DNase I in the presence of 10 mM MgCl₂ at 37°C for 1 h to eliminate any residual contaminating plasmids from the transfection. Infection of CEMx174 or MT4 cells was performed by incubating 10⁶ cells at 37°C for 2 h with an amount of virus equivalent to 10 ng of p27 antigen. Infected cells were then washed extensively with PBS and resuspended in fresh medium. Cells were split at a 1:3 ratio twice per week if they had grown to a sufficient level; otherwise the culture fluid was replaced with fresh medium. Supernatants were monitored for virus production by SIV p27 antigen capture assay using a Coulter SIV core antigen assay kit. Virus replication was also performed in primary rhesus monkey peripheral blood mononuclear cells (PBMCs). Activated PBMCs (5 × 10⁶) were infected with SIV stocks containing 10 ng of p27 at 37°C for 2 h; the cells were then washed extensively to remove any remaining virus. Cells were maintained in RPMI 1640 medium supplemented with 10% heat-inactivated fetal bovine serum and interleukin-2 (20 U/ml). Virus production in culture fluids was monitored by SIV p27 antigen capture assay using a Coulter SIV core antigen capture kit.

Detection of viral DNA.

At various times postinfection, cells were collected and washed with PBS. Cellular DNA was isolated using a QIAamp DNA mini kit (Qiagen). DNA samples were analyzed by PCR using primers sg1 and sg2 to amplify a 114-bp fragment in the gag gene. PCR was performed with 0.1 to 1 μg of sample DNA, 50 mM Tris-HCl (pH 8.0), 50 mM KCl, 1.5 mM MgCl₂, 2.5 U of Taq polymerase, 0.2 mM deoxynucleoside triphosphates, 20 pmol of reverse primer, and 20 pmol of forward primer as follows: 95°C for 3 min; 25 cycles at 94°C for 30 s, 55°C for 30 s, and 72°C for 1 min; and 72°C for 10 min. Products were separated on 2% agarose gels. In the case of transient infections, cells were exposed to virus as described above, collected after 6 h, and washed extensively with PBS; negative infection controls for each construct were performed at 4°C using precooled cells and viruses. Cellular DNA was isolated using a QIAamp DNA mini kit (Qiagen). DNA samples were analyzed by PCR as described above except that the sg1 primer was ³²P labeled and reactions were standardized by simultaneous amplification of a 567-bp DNA fragment of the human β-actin gene as an internal control as described previously (20).

RESULTS

Generation of simplified SIV constructs.

A series of SIV mutants containing deletions within various nonstructural genes was constructed as described in Materials and Methods (Fig. 1A). In all mutants, both the vpx and vpr genes were completely deleted, while most of the vif gene was deleted; (only the region encoding the first 21 amino acids of Vif remained). The nef gene was interrupted by deletion of the sequences from positions 9500 to 9674. In the case of the Δ5 construct, only one nonstructural gene, rev, was retained, and the tat gene was inactivated by a large deletion that included the first 145 bp of the first tat exon. In all constructs that contained only structural genes (Δ6 series), both the tat and rev genes were inactivated by deletion of the first exons of tat and rev including their splice donor sites.

To compensate for removal of the Rev/RRE posttranscriptional regulation system, a 173-bp CTE sequence of SRV-1 was inserted into the site of the nef deletion to form the construct termed Δ6-CTE. To increase the expression of viral genes in the absence of the tat gene, a CMV immediate-early (IE) promoter (including its enhancer and TATA box) were inserted into the LTR U3 promoter region of SIV (a 473-bp fragment from its upstream sequence to the TATA box) to generate construct Δ6-CTE-CMV (Δ6CC). To determine whether this chimeric LTR was functional, it was inserted into a wild-type construct containing the nef deletion. This construct, termed SIVmac239Δnef-CMV (Fig. 1B), was transfected into COS-7 cells to produce viral stock. As shown in Fig. 2, SIVmac239Δnef-CMV replicated similarly to wild-type virus in CEMx174 cells. This result confirms that the chimeric LTR with the CMV promoter can be used efficiently by SIV in the CEMx174 cell line.

FIG. 2.

Growth curves of viruses containing a chimeric CMV-LTR. Equivalent amounts of viruses were used to infect CEMx174 cells. Viral replication was monitored by RT assay of culture fluids. Mock infection denotes exposure of cells to heat-inactivated wild-type (WT) virus as a negative control.

Deletion of the upstream sequences of the env gene may inhibit the efficient translation of envelope proteins due to deletion of upstream sequences that include splice acceptor sites. Therefore, an IRES element was inserted between the gag-pol and envelope genes into the Δ6CCI construct to aid expression. To confirm that the IRES element was functional in the SIV construct, the envelope gene (from its ATG to the SacI site) was replaced with the EGFP reporter gene. The EGFP gene is in the same open reading frame as the env gene; this construct, termed Δ6CCI-GFP (Fig. 1C), was transfected into COS-7 cells. Both fluorescence microscopy (not shown) and fluorescein isothiocyanate-gated fluorescence-activated cell sorting confirmed the expression of the EGFP gene. The results of Fig. 3 show that the IRES of poliovirus resulted in the highest degree of expression of GFP (51%), whereas the ECMV IRES resulted in only 30% expression of GFP, while background levels were about 10%. Similar results were obtained in each of three separate experiments. Therefore, in all other experiments we used the variant of the Δ6CCI construct that contained the poliovirus IRES (Δ6CCI-P).

FIG. 3.

Fluorescence isothiocyanate(FITC)-gated fluorescence-activated cell sorting analysis of simplified SIV vectors. A SIV-Δ6-CTE-CMV-IRES-GFP vector containing the IRES of EMCV (construct Δ6CCI-E-GFP) or of poliovirus (construct Δ6CCI-P-GFP) was transfected into COS-7, and the expression of GFP was analyzed by flow cytometry. The percentage of GFP-positive cells is indicated in each graph. The x axis designates cell number, while the y axis refers to the fluorescence density of GFP. Mock denotes transfection of COS-7 cells by the Δ6CCI construct, which lacks the GFP gene, as a negative control; hence, the background of fluorescence in these studies was about 10%.

Production of modified SIV.

All of these simplified SIV constructs were transfected into COS-7 cells, and virus production in supernatants was detected by RT assay and p27 antigen quantification. As shown in Fig. 4, viruses without tat (Δ5) were efficiently produced after transfection of COS-7 cell lines. Interestingly, mutants with deletions in all six nonstructural genes (Δ6) were produced at levels 100 times less than those of wild-type virus in the absence of additional elements. In contrast, addition of the CTE (construct Δ6-CTE), the chimeric CMV promoter (construct Δ6CC), and the IRES (construct Δ6CCI), efficiently increased SIV production to levels comparable to those of wild-type virus.

FIG. 4.

Virus production following transfection of COS-7 cells. SIV wild-type (WT) or mutant constructs were transfected into COS-7 cells. Levels of RT activity and SIV p27 antigen in culture fluids were quantified at 60 h after transfection and plotted. Results were calculated on the basis of three independent transfections and are shown as averages ± standard deviations.

We also analyzed viral protein production by radiolabeling and immunoprecipitation as described in Materials and Methods. Figure 5 presents the viral protein pattern of viruses produced during 1 h by transfected COS-7 cells; the data show that the Δ5 construct was able to produce viral proteins efficiently, while the Δ6 virus was severely impaired in this regard. These findings are similar to those obtained through use of the p27 enzyme-linked immunosorbent assay and RT assay (Fig. 4). In contrast, viral proteins were efficiently produced with the help of the CTE (construct Δ6-CTE), the chimeric CMV promoter (construct Δ6CC), and the IRES (construct Δ6CCI). However, these simplified viruses were devoid of some proteins such as those between bands p6 and p15; these may represent accessory proteins such as Vpr. Immunoprecipitation of viral proteins in cell lysates with MAbs against SIV p27 showed that Gag protein was efficiently expressed in these simplified constructs (although certainly not coprecipitated with Vpr). However, the processing of Gag precursor proteins was delayed, resulting in accumulation of Pr55 (Fig. 6). Immunoprecipitation of viral proteins in cell lysates with MAbs against SIV gp120 antigen showed that viral gp120 was efficiently expressed only in the case of constructs Δ5 and Δ6CCI; expression of gp120 protein in the Δ6, Δ6-CTE, and Δ6CC constructs was diminished (Fig. 6).

FIG. 5.

Protein patterns of viral particles. ³⁵S-labeled viral progeny that were released during 1 h by transfected COS-7 cells were purified at 24 h after transfection. Proteins were analyzed by PAGE. Mock denotes transfection of COS-7 cells by vector pSP73, not containing any SIV genomic material, as a negative control. WT, wild type.

FIG. 6.

Identification of SIV gp120 and p27 antigens. COS-7 cells transfected with wild-type (WT) or mutated SIV constructs were radiolabeled with both [³⁵S]Met and [³⁵S]Cys, and viral proteins in cell lysates were then immunoprecipitated with MAbs against SIV gp120 or p27 antigen. The bands of gp120 and Gag proteins are shown. Mock denotes transfection of cells by vector pSP73 as a negative control.

Viral production was also analyzed by electron microscopy. Figure 7 shows that these simplified viruses retained morphology typical of wild-type mature virions in the cases of constructs Δ5, Δ6-CTE, and Δ6CC, indicating that these viruses retained the ability to form structures that appear to have both envelopes and normally dense cores. The Δ6 viruses could not be analyzed by electron microscopy because of very low levels of particle formation and production.

FIG. 7.

Morphology of virions produced by wild-type (WT) or simplified SIV constructs. COS-7 cells transfected with wild-type or simplified SIV constructs were fixed, sectioned, and stained at 40 h posttransfection and then visualized by electron microscopy. The bar represents 100 nm. Mature virions are indicated by the arrows.

Deletion of accessory genes affects both the specificity and efficiency of SIV genomic RNA packaging.

Inactivation of the rev gene may impair viral RNA packaging, because Rev regulates viral RNA export from the nucleus as well as the expression of structural proteins. Therefore, we investigated the extent to which our simplified viruses could package viral RNA by RT-PCR. For this purpose, two pairs of primers were used; one of these amplified total viral RNA, while the other amplified only full-length genomic RNA. As shown in Fig. 8, the Δ6 viruses were able to package viral genomic RNA to extents of only about 20% of that of total viral RNA and about 50% of viral genomic RNA packaged by wild-type viruses. These results indicate that both the efficiency and specificity of viral genomic RNA packaging were significantly diminished in the Δ6 virus that contained only viral structural genes. Insertion of the CTE and the CMV promoter increased the specificity but not the efficiency of viral genomic RNA packaging, since the Δ6-CTE and Δ6CC constructs packaged viral genomic RNA with specificities of 75 and 97%, respectively. Interestingly, the additional insertion of the IRES remarkably increased the efficiency of viral genomic RNA packaging (Fig. 8, construct Δ6CCI). The simplified construct, Δ6CCI-P, was able to specifically package viral genomic RNA as efficiently as wild-type virus.

FIG. 8.

Packaging of viral RNA as assessed by RT-PCR. RNA was purified from viruses derived from transfected COS-7 cells. Equivalent amounts of virus, based on levels of p27 antigen, were used as template in quantitative RT-PCR to detect the presence of total viral RNA (A) and of the full-length viral RNA genome (B) in 18-cycle PCRs. Reactions run with RNA template that had been digested by DNase-free RNase served as a negative control for each sample to exclude any potential DNA contamination. Relative amounts of a 114-bp DNA product representing full-length viral RNA (B) were quantified by molecular imaging, with wild-type (WT) levels arbitrarily set at 1.0, to determine the efficiency of genomic RNA packaging. The relative amounts of full-length viral RNA (B) to total viral RNA (A) in each sample were also quantified to determine the specificity of viral RNA packaging. The relative amounts of viral RNA that were packaged were determined on the basis of four different experiments and are shown as averages ± standard deviations.

Infection of CEMx174 cells.

We next investigated the infectiousness and replication capacity of these simplified viruses. Virus stocks were used to infect CEMx174 cells as described above, and culture fluids were monitored for viral replication by SIV p27 antigen capture assay. As shown in Fig. 9A, detectable amounts of viral antigen were detected only after 6 days in the case of Δ6CCI-P; the other simplified constructs showed no signs of replication in these studies. The positive p27 result for the Δ6CCI-P construct was seen with duplicate experiments. To further confirm this finding, proviral DNA was harvested from cells at various times after infection and subjected to PCR analysis. The data in Fig. 9B show that infection by Δ6CCI-P virus of CEMx174 cells had indeed occurred. However, long-term culture of these infected cells over 6 months did not show any signs of reverted or more replication-competent viruses (data not shown). Thus, this simplified SIV is extremely attenuated in replication capacity yet can still infect CEMx174 cells.

FIG. 9.

Detection of viral DNA and p27 antigen after infection of CEMx174 cells. (A) Equivalent amounts of wild-type (WT) or modified viruses, based on p27 content, were used to infect CEMx174 cells. Viral replication was monitored by p27 antigen assay of culture fluids. Mock infection denotes exposure of cells to heat-inactivated wild-type virus as a negative control. The dashed line, representing 0.01 ng of p27 per ml, indicates the threshold sensitivity of the assay. (B) At various times postinfection, cellular DNA was analyzed by PCR using primers sg1 and sg2 to amplify a 114-bp fragment in the gag region (20). PCR products were separated on 2% agarose gels. Lane 1, in- fection by wild-type virus after 4 days; lane 2, infection by heat-inactivated wild-type virus after 4 days; lanes 3 to 6, infection by Δ6CCI-P virus at days 4, 7, 14, and 21, respectively, after infection; lane 7, infection by heat-inactivated Δ6CCI-P virus after 4 days. M designates 100-bp ladder. (C) PCR analysis of viral DNA in transiently infected CEMx174 cells, as described in Materials and Methods. The 114-bp band of viral DNA and the 567-bp band of β-actin cellular DNA used as an internal control are indicated. Infections performed and maintained at 4°C served as negative controls for each of the constructs.

To further investigate this subject, we also performed transient infections of CEMx174 cells alongside control experiments performed at 4°C. The PCR results in Fig. 9C show that the Δ6CCI-P virus was indeed able to infect CEMx174 cells but less well than wild-type virus. All infections performed and maintained at 4°C yielded negative results.

Continuous propagation of simplified SIVs in MT4 cells.

In addition to the simplified construct Δ6CCI-P, we also generated five additional viruses constructs in which the Rev/RRE system or vif gene was maintained, i.e., Δ5CCI, Δ4, Δ4-CMV, Δ5-CMV, and Δ2-CMV (Materials and Methods; Fig. 1). The results of infections of CEMx174 cells are shown in Fig. 10A. The Δ4 and Δ5 mutants showed no sign of replication, while Δ5CCI, Δ4-CMV, Δ5-CMV, and Δ2-CMV replicated with similarly impaired efficiency as the Δ6CCI-P construct (Fig. 9A). We further infected MT4 cells, which have been shown to be permissive for replication of either vif- or tat-negative SIVmac239 mutants (8, 37, 54). Remarkably, the viruses that contained the CMV promoter, Δ4-CMV and Δ2-CMV, showed efficient although delayed replication kinetics in MT4 cells. The Δ5-CMV virus yielded persistent low-level replication in MT4 cells, while results for the Δ6CCI-P and Δ5CCI viruses were similar in MT4 and CEMx174 cells (Fig. 10B).

FIG. 10.

Replication capacity of the Δ4 and Δ5 constructs. Equivalent amounts of wild-type (WT) or modified viruses, based on p27 content, were used to infect both CEMx174 and MT4 cells. Viral replication was monitored by p27 antigen assay of culture fluids. Mock infection denotes exposure of cells to heat-inactivated wild-type virus as a negative control. The dashed line, representing 0.01 ng of p27 per ml, indicates the threshold sensitivity of the assay. (A) Growth curve in CEMx174 cells. (B) Growth curves in MT4 cells. (C) Growth curves of second-passage MT4-derived viruses, from the experiment in panel B, in fresh MT4 cells.

Cell-free viruses harvested after initial infection of MT4 cells were then passaged in this same cell line. As shown in Fig. 10C, the Δ2-CMV, Δ4-CMV, and Δ5-CMV viruses all replicated similarly as in the initial infections. These results demonstrate that these three simplified viruses are all stably attenuated in vitro.

Infection of monkey PBMCs.

We also investigated the infectiousness of our simplified viruses in monkey PBMCs, using protocols described previously (20). As shown in Fig. 11A, the Δ2-CMV and Δ4-CMV viruses displayed transient replication capacity in these cells, while the Δ5-CMV, Δ6CCI-P, and Δ5CCI viruses showed no sign of replication. We further assessed the presence of viral DNA in monkey PBMCs by PCR using primers sg1 and sg2. The results in Fig. 11B show that our simplified viruses were indeed able to infect monkey PBMCs, albeit at low efficiency.

FIG. 11.

Infection of monkey PBMCs. (A) Equivalent amounts of wild-type (WT) or modified viruses, based on p27 content, were used to infect monkey PBMCs. Viral replication was monitored by p27 antigen assay of culture fluids. Mock infection denotes exposure of cells to heat-inactivated wild-type virus as a negative control. The dashed line, representing 0.01 ng of p27 per ml, indicates the threshold sensitivity of the assay. (B) PCR analysis of viral DNA in transiently infected monkey PBMCs as described in Materials and Methods. The 114-bp band of viral DNA is indicated. Infections performed and maintained at 4°C served as negative controls for each of the constructs.

DISCUSSION

We have generated a series of simplified SIVmac constructs that are devoid of several or all accessory genes. One of these, termed Δ6CCI, with the help of a CTE, the CMV promoter, and an IRES, can efficiently produce mature virions that package viral genomic RNA as well as do wild-type viruses. These viruses also retain the ability to infect target cells yet are deficient in replication capacity. The Δ6CCI construct might be suitable for use as a DNA vaccine, because it causes expression of natural viral antigens that are exposed during infection (27).

Highly attenuated SIV mutants containing partial deletions in nonessential genes have elicited strong protection against pathogenic challenge (1, 12, 14, 22, 23, 50). Wide ranges of attenuation levels have been achieved in such experiments, and protective efficiency was shown to be inversely proportional to the degree of attenuation (23). Prevailing opinion suggests that live attenuated strategies may fail if viruses are too attenuated (41). However, a highly attenuated SIV lacking nef, vpr, vpx, and upstream sequences in U3 (SIVmac239Δ4) maintained ability to induce reasonable levels of protection against vaginal challenge (14, 23). A SIVmac239Δvif construct, which could grow consistently only on vif-complementing cells, was able to infect rhesus monkeys and to elicit persistent, albeit weak, immune responses (23). Thus, even severely attenuated viruses retain ability to induce protective immune responses, something that no other vaccine strategy has been shown to accomplish. We believe that it is worthwhile to further attenuate viruses such as SIV until they are devoid of disease-causing ability and to then increase their capacity to elicit protective immunity through improved immunization protocols (46). As an example, a simplified bovine leukemia virus has been successfully generated, and in vivo studies have shown that it is both immunogenic and safe and can induce protective immune responses against wild-type viruses in rabbits (3, 4, 5). As shown here, we have constructed simplified forms of SIV that might now be studied in primate models.

Toward this end, we eliminated all of the nonstructural genes of SIV through large deletions and introduced three functional elements in their stead to restore viral production. First, a 173-bp CTE of SRV-1 was used to increase viral genomic RNA transportation, because this element has been proven competent to compensate for deficits of the Rev/RRE posttranscriptional regulation system (45, 48, 53). These findings are confirmed by our insertion of the CTE into the Δ6 construct, resulting in increased expression of viral structural proteins (Δ6-CTE [Fig. 4 and 5]).

Tat is essential for the replication of both HIV and SIV. We therefore used a strong promoter, the CMV IE promoter, to increase the efficiency of transcription and to partially compensate for the deletion of tat, since previous work has shown the rationale for this approach (8). We found that a chimeric CMV-LTR, when introduced into the Δ6-CTE construct, significantly increased the expression of viral structural genes (Δ6CC [Fig. 4 and 5]). Remarkably, this CMV-LTR can also drive the efficient replication of the Δ2-CMV, Δ4-CMV, and Δ5-CMV mutants in MT4 cells (Fig. 10B and C).

Although vif has been shown to be essential for replication of both HIV-1 and SIV, several groups have suggested that this effect may be cell type dependent. In the case of CEMx174 cells, vif-deficient SIVmac239 mutants were able to establish productive infection (54). Others, however, have suggested that replication of vif-mutated SIVmac239 viruses in CEMx174 cells was severely restricted (37). Gibbs et al. showed that Δ5 (vif-deficient) mutants replicated to high levels after prolonged culture in CEMx174 cells, while other vif-deficient viruses displayed only low-level replication in this same cell line (18). Our data are similar in that a construct that retained the vif gene, Δ5CCI, showed similar replication patterns in both CEMx174 and MT4 cells as did the vif-deficient Δ6CCI virus. Therefore, replication of a SIVmac239 mutant that lacks vif can occur, albeit with impairment, in CEMx174 cells.

Our Δ6CCI mutant is at least as attenuated as the Δ5 (vif-deficient) virus of Gibbs et al. (18), but the additional removal of both tat and rev, which are important in the pathogenesis of HIV-1 (17, 30, 38, 47, 52), may provide an extra margin of safety. It has been shown that vaccination with proviral DNA that encodes intact but noninfectious viruses may induce a protective immune response (49). Our Δ6CCI construct retains the ability to produce all viral structural proteins, to form mature virions, and to transiently infect target cells while being severely impaired in regard to replication. These properties make it a good DNA vaccine candidate, since conformational epitopes that are exposed only during infection are believed to elicit cross-subtype immune responsiveness (27).

In the case of HIV-1, vif-defective viruses have been shown to persistently replicate in primary macrophages (9) and were able to enter PBMCs with the same efficiency as wild-type virus (11). The fact that vif-negative SIV could induce antibody response in macaques after a single injection suggests that these viruses were able to complete at least a single round of infection in vivo. Theoretically, our Δ6CCI construct should also retain this ability. At the same time, the deficiency of propagation of our Δ6CCI construct seen in primary cells might not compromise its utility as a proviral DNA vaccine.

Our large deletions had removed sequences between the gag-pol and env genes, including splice acceptor sites for env, therefore diminishing the translation of the latter gene (Fig. 6). This was corrected by introduction of a functional poliovirus-derived IRES between the gag-pol and env genes; the result was that env gene expression was rendered independent of splicing and dependent on the same mRNA as that involved in expression of the gag-pol gene. The presence of this IRES in the Δ6CC construct significantly increased the expression of viral structural proteins and especially that of Env (Fig. 6, Δ6CCI). Furthermore, the Δ6CCI construct produced viral particles with comparable efficiency to wild-type SIV constructs. However, this simplified SIV, which contained only viral structural genes, was extremely attenuated in replication capacity in CEMx174 cells (Fig. 9).

Our simpler SIVs differ from other, partially deleted SIV constructs (18) in that they contain only structural genes. These simplified SIVs are live but diminished in replication capacity and are presumably attenuated due to the deletion of nonstructural genes and the functional loss of these regulatory elements. Although few mechanistic studies have been performed on viruses of this type, it is known that HIV that was inactivated in regard to the Rev/RRE system, by replacement of rev with a CTE, regained replication competence while displaying deficient processing of the Gag precursor protein Pr55 (53). Similar results have now been observed with our simplified SIVs. However, this effect was not due solely to abrogation of the Rev/RRE system, since our Δ5 construct, which retains the Rev/RRE, displayed similar patterns of deficiency (Fig. 6). Furthermore, an even simpler construct, Δ6, was deficient in both specificity and efficiency of encapsidation of viral genomic RNA. We found that replacement of rev by the CTE partially compensated for this impairment in specificity and that introduction of a stronger promoter, i.e., the CMV IE promoter even further increased the specificity of packaging. These results indicate that packaging of viral genomic RNA may require efficiency in regard to both transcription and transport of viral genomic RNA. The fact that insertion of an IRES, i.e., construct Δ6CCI, even further increased the efficiency of packaging also indicates that both the expression and length of viral genomic RNA may be key factors in this regard.

However, if these simplified viruses are to be used as live attenuated vaccines, further modifications may be required to improve their replication capacity in order to efficiently generate protective immunity (23). In this context, our simplified SIV constructs still contain certain trans-activated elements within the LTR, such as TAR sequences, that may affect SIV replication in the absence of Tat. Previous work has shown that the strong CMV promoter might only partially correct for the absence of Tat protein (8). Our results also show that the CMV promoter can rescue our tat-negative viruses in MT4 cells but not in CEMx174 cells. We are planning to generate SIV constructs that contain LTRs of simpler retroviruses to create viruses that are fully independent of retroviral trans-activated factors (3, 7, 46). We next hope to evaluate the infectivity, safety, immunogenicity, and protective ability of our constructs in macaque monkeys by inoculation of viral DNA constructs.