Sequences Just Upstream of the Simian Immunodeficiency Virus Core Enhancer Allow Efficient Replication in the Absence of NF-κB and Sp1 Binding Elements

Construction of LTR mutants.

An overview of the LTR mutants analyzed is given in Fig. 2. LTR mutations were introduced into the proviral clone SIVmac239 ΔNU, in which 182 bp in the nef-unique region and 334 bp of the upstream U3 region are deleted (11). The 334-bp U3 deletion affects only sequences that were selectively deleted in macaques infected with nef-defective SIVmac239 (22). First, site-directed mutagenesis was performed by spliced overlap extension PCR to generate unique SstII and XhoI sites at the 5′ end of the US65 region (ΔUSK-mutant). PCR products were gel purified and inserted into the SIVmac239 ΔNU clone by using the unique XmaI and EcoRI sites in the nef-unique region and in the vector sequences flanking the 3′ end of the provirus. Subsequently, the mutated 3′ LTR was amplified with primers pSphI (5′-CATTAAAGCTTGTCGACTGGAAGGGATTTAT-3′) and pSphII (5′-AATTGGCGCCAATCTGCTAGGGATTTTCCTG-3′) and inserted into the HindIII site located upstream of the 5′ end of the provirus and the NarI site located at the 3′ end of the 5′ LTR. Cloning of this fragment generated a unique SalI site upstream of the 5′ end of the provirus. All point mutants (simian factor 2 [SF2], SF3, and SF123 binding sites) and most deletion mutants were also generated by spliced overlap extension PCR. Mutations were first introduced into the 3′ LTR. Subsequently, the mutated 3′ LTRs were PCR amplified and inserted into the unique SalI and NarI sites, to replace the 5′ LTR. Mutations were always introduced into both LTRs to prevent recombination. SIVmac239 LTR mutants containing deletions in the upstream U3 region in conjunction with deletions in the core enhancer were generated by using the same protocol except that templates with deletions in the NF-κB and/or Sp1 binding sites were used in the first round of PCR. The ΔUS400 mutant was created by using a 5′ primer containing the XmaI site, the polypurine tract, and the NF-κB binding site. All PCR-derived inserts were sequenced to confirm that only the intended changes were present.

Transactivation assays.

The mutated LTRs (Fig. 2) were PCR amplified and cloned upstream of the luciferase reporter gene. For Tat expression, the first coding exon of the SIVmac239 tat gene was cloned into an expression plasmid containing a cytomegalovirus promoter (pcDNAIII; Invitrogen, San Diego, Calif.). To assess the promoter activities, COS-1 cells were seeded at a fivefold dilution and transfected with 1 μg of the various SIVmac LTR-luciferase plasmids, or cotransfected with 1 μg of a SIVmac Tat expression plasmid, by a DEAE-dextran method on the following day (3). The cells were harvested 48 h after transfection, and luciferase activity was measured with a commercially available kit as recommended by the manufacturer (Promega, Madison, Wis.). To allow a better comparison among individual experiments, stock DNA preparations from an SIVmac239 wild-type (239wt) LTR control plasmid were always included as a positive control. The amount of luciferase activity obtained after transfection of 239wt LTR plasmid was considered 100% activity, and the activities of the mutated LTRs were determined relative to this value. Furthermore, 1 μg of a plasmid expressing the β-galactosidase gene under control of the cytomegalovirus promoter was cotransfected as an internal control for transfection efficiency. The luciferase values were normalized for protein content and β-galactosidase activity, which were assayed by using commercially available kits (Promega).

Production of virus stocks.

For virus production, semiconfluent COS-1 cells were transfected with 5 μg of full-length proviral DNA constructs as described above. At 48 h posttransfection, the medium was replaced with fresh Dulbecco modified Eagle medium supplemented with 10% fetal calf serum (FCS). Virus stocks were harvested at 4 days posttransfection from cell culture supernatant, passed through a 0.45-μm-pore-size filter, and stored at −80°C. To produce virus stocks from CEMx174 cells, the cultures were diluted with fresh medium and transfected with 3 μg of proviral DNA the following day, using a DEAE-dextran protocol (29). When cytopathic effects were observed, the cells were mixed with an equal volume of uninfected cells, pelleted, resuspended in fresh medium, and cultivated overnight. Thereafter, cells were pelleted again and the cell-free culture supernatant was filtered and stored in aliquots at −80°C. The concentration of p27 antigen was determined with a commercially available antigen capture kit (Innogenetics, Zwijndrecht, Belgium). The viral stocks were used to infect the human T-B hybrid cell line CEMx174, rPBMC, and herpesvirus saimiri-transformed T-cell lines.

Cell culture.

CEMx174 and COS-1 cells were maintained as described previously (13). Herpesvirus saimiri-transformed T-cell lines of human (CB15) and of rhesus macaque (MmHF7062A) origin were kept in a mixture of equal amounts of RPMI 1640 and GC medium (Vitromex, Vilshofen, Germany) supplemented with 10% FCS and 100 U of interleukin-2 per ml. rPBMC, obtained from healthy rhesus macaques that were seronegative for SIV, type D retroviruses, and simian foamy virus, were isolated by using lymphocyte separation medium (Organon Teknika Corporation, Durham, N.C.), stimulated for 2 to 3 days with 5 μg of phytohemagglutinin (PHA) per ml, and cultured in RPMI 1640 medium with 20% FCS and 50 U of interleukin-2 per ml.

Viral replication.

CEMx174, CB15, and MmHF7062A cells split 1:2 to 1:5 the previous day were infected with an aliquot of SIVmac containing 2 ng of viral p27 antigen. Activated rPBMC were infected with the same amount of SIVmac upon PHA stimulation as described above. Cell culture supernatants were sampled at regular intervals and stored at −80°C, and virus production was measured by reverse transcriptase assay as described elsewhere (31).

In vitro transcription and translation reactions.

Elf-1 and Ets-1 expression plasmids (39) were kindly provided by D. M. Markovitz and J. M. Leiden. The in vitro transcription and translation reactions were performed with a commercially available kit (Promega) as instructed by the manufacturer.

EMSAs.

Electrophoretic mobility shift assays (EMSAs) were carried out with double-stranded oligonucleotide probes. Complementary single-stranded oligonucleotides were custom made (Eurogentec, Seraing, Belgium), and double-stranded oligonucleotides were generated by annealing of complementary oligonucleotides. Prior to the annealing reaction, one oligonucleotide strand was [γ-³²P]ATP end labeled by polynucleotide kinase (New England Biolabs, Schwalbach, Germany). Double-stranded oligonucleotides were purified by electrophoresis in 12% native polyacrylamide gels and eluted from the gel matrix in Tris-EDTA buffer (pH 8.0) overnight at 4°C. The following oligonucleotide probes were used (the sequences of the sense strands are shown in 5′→3′ orientation): HIV-2 PuB2 (CAGCTATACTTGGTCAGGGCAGGAAGTAACTA), murine sarcoma virus (MSV) (TGCGCGCTTCCGCTCTCCGAG), SIV PuB1 (TGTCAGAGGAAGAGGTTAG), SIV PuB2 (TGGCTGACAAGAAGGAAACTCGCTGAAACAGCA), mut-SF3 (TGGCTGACAAGAAGGGAGCTCGCTGAAACAGCA), mut-SF2 (TGGCTGACATATGGGAAACTCGCTGAAACAGCA), and mut-Elf (TGGCTGACAAGAGTTGAGCTCGCTGAAACAGCA (mutated positions are underlined). DNA-protein binding reactions using T-cell nuclear extract were carried out in 75 mM KCl–10 mM Tris (pH 7.5)–1 mM dithiothreitol (DTT)–1 mM EDTA–4% Ficoll and contained 20,000 cpm of labeled oligonucleotide probe, 10 μg of nuclear extract, 250 ng of double-stranded poly(dI-dC) (Pharmacia, Freiburg, Germany), and 250 ng of calf thymus DNA (Merck, Darmstadt, Germany) in a final volume of 20 μl. DNA-protein binding reactions using in vitro-translated Ets-1 and Elf-1 proteins were carried out in 75 mM KCl–10 mM Tris (pH 7.5)–1 mM DTT–1 mM EDTA–4% Ficoll and contained 20,000 cpm of labeled oligonucleotide probe, 6 μl of in vitro-translated protein, and 380 ng of double-stranded poly(dI-dC) in a final volume of 30 μl. For supershift analysis, the reaction mixture additionally contained 2 μl of Elf-1 antibody (Santa Cruz Biotechnology, Santa Cruz, Calif.). The reaction mixtures were incubated for 30 min at room temperature and subsequently separated in 4% nondenaturing polyacrylamide gels. The gels contained 0.25 mM KCl and 0.5 mM DTT and were run in 0.25× Tris-borate-EDTA buffer at 120 V for 4 h at 4°C. Immunoblots were probed with a 1:10,000 dilution of the Elf-1 antibody in conjunction with a 1:5,000 dilution of the commercially available anti-rabbit horseradish peroxidase-coupled antibody (Dianova, Hamburg, Germany). Immunoblots were developed by using an enhanced chemiluminescence detection system (Amersham, Chicago, Ill.) as specified by the manufacturer.

Nuclear extracts.

MmHF7062A cells (1 × 10⁷ to 5 × 10⁷) were lysed with 10 mM HEPES (pH 7.9)–1.5 mM MgCl₂–10 mM KCl–0.5 mM DTT–0.5 mM phenylmethylsulfonyl fluoride (PMSF)–0.1% (vol/vol) Nonidet P-40. Nuclei were pelleted by centrifugation in a microcentrifuge (Biofuge 13; Heracus Sepratech) at 3,000 rpm for 15 min at 4°C and resuspended in 20 mM HEPES (pH 7.9)–0.42 M NaCl–1.5 mM MgCl₂–0.2 mM EDTA–0.5 mM DTT–0.5 mM PMSF–8 μg of aprotinin per ml–2 μg of leupeptin per ml–25% (vol/vol) glycerol. Reaction mixtures were incubated for 15 min at 4°C. Nuclear debris was pelleted by centrifugation in a microcentrifuge (Biofuge 13) at 14,000 rpm for 15 min at 4°C, and the cleared supernatants were dialyzed for 4 h at 4°C against 20 mM HEPES (pH 7.9)–0.1 M KCl–0.2 mM EDTA–0.5 mM DTT–0.5 mM PMSF–20% (vol/vol) glycerol. Aliquots were stored at −80°C, and protein concentrations were assayed by using a commercial kit (Bio-Rad, Munich, Germany).

Transcriptional activities of mutated SIV LTRs.

To investigate the influence of deletions and point mutations in the US region and the NF-κB and Sp1 binding sites on transcriptional activity, the luciferase reporter gene was expressed under the control of the mutated SIVmac239 LTRs in the presence or absence of Tat (Fig. 3). Removal of the 334 bp of US sequences that were selectively deleted in vivo had no significant effect (ΔNU, 111% ± 20% [Fig. 3]). Additional deletions of all U3 sequences upstream of the NF-κB binding site, except those important for integration, resulted in only very moderately decreased activity (ΔUS400, 77% ± 19% [Fig. 3]). Similar marginal effects were observed when the previously described factor binding sites SF1 to SF3 were mutated (mut-SF123, 67% ± 10% [Fig. 3]). In agreement with previously published data (14), deletion of the single NF-κB site had no significant effect (92% ± 43%), whereas deletion of all Sp1 binding sites resulted in an approximately threefold decrease of transcriptional activity in the presence of Tat (31% ± 2% [Fig. 3]). In the absence of the Sp1 sites, further deletion of 50 bp of the US65 region reduced reporter gene activity in the presence of Tat about two- to threefold, to approximately 15% of the 239wt LTR activity (ΔSp1/US384, 14% ± 8% [Fig. 3]). Even in the absence of the entire 400 bp of US sequences and all NF-κB and Sp1 binding sites, the truncated SIV promoter maintained approximately 15% of 239wt LTR activity and was responsive to Tat activation (ΔNF-κB/Sp1/US400, 17% ± 3% [Fig. 3]). Thus, under the experimental conditions used, deletion of about 87% of the SIVmac U3 region (452 bp), including the core enhancer and most of the US sequences, resulted in only a sixfold decrease of promoter activity and had little effect on Tat transactivation in COS-1 cells. Similar results were obtained after cotransfection of Jurkat T cells with a Tat expression vector and a luciferase reporter plasmid (ΔNU, 131% ± 34%; ΔUS400, 70% ± 43%; ΔNF-κB/US384, 79% ± 16%; ΔSp1/US384, 24% ± 13%; ΔNF-κB/Sp1, 46% ± 20%; ΔNF-κB/Sp1/US384, 19% ± 12%; numbers represent percentages of wild-type reporter activity obtained from six transfections). The absolute values were lower and more variable, however, than those obtained for COS-1 cells.

Virus production in COS-1 cells.

Full-length proviral SIVmac239 constructs, containing mutations in both LTRs to prevent recombination, were transfected into COS-1 cells for virus production. A defective nef gene resulted in an approximately threefold reduction of p27 core antigen production compared to nef-open SIVmac239 (nef*, 31% ± 2.8% [Fig. 4A]). The deletion of 354 bp of upstream U3 sequences had no significant effect (ΔUS354, 29% ± 0.3% [Fig. 4A]). Further deletion of all U3 sequences upstream of the NF-κB site, except those required for integration, reduced p27 production about threefold compared to SIVmac239 nef* (ΔUS400, 11% ± 1.9% [Fig. 4A]). Mutating the previously described SF1 to SF3 binding sites in the SIVmac239 ΔNU variant resulted in an approximately twofold decrease of p27 production compared to nef-defective SIVmac239 (mut-SF123, 14% ± 0.3% [Fig. 4A]). Thus, deletion and mutation of the SF1 to SF3 sites had comparable effects on p27 production (ΔUS384, 15% ± 2.5%; mut-SF123, 14% ± 0.3% [Fig. 4A]).

In the presence of the NF-κB and Sp1 binding sites, deletion of the 60 bp just upstream of the core enhancer reduced virus production about threefold (Fig. 4A). To assess the relevance of the US region for virus production in the absence of core enhancer elements, SIVmac239 LTR variants containing changes in both the US region and the core enhancer elements were also investigated (Fig. 4B). Deletion of the Sp1 sites reduced p27 production about 40-fold (2.5% ± 1.8%), whereas removal of the single NF-κB site had only a 2-fold effect (45% ± 15%) (Fig. 4B). In the absence of the NF-κB site, an additional deletion of 374 bp of US sequences had little effect, and virus production was comparable to that of nef-defective SIVmac239 (ΔNF-κB/US374, 23% ± 5%, and nef*, 31% ± 2.8% of the value for nef-open SIVmac [Fig. 4]). In contrast, deletion of these US sequences resulted in approximately 15-fold-lower p27 levels in the absence of the four Sp1 sites (ΔSp1, 2.5% ± 1.8%; ΔSp1/US374, 0.17% ± 0.13%). Deletion of the region encompassing the SF2 and SF3 binding sites (US384) had a four- to sixfold negative effect on p27 antigen production in the absence of NF-κB or Sp1 binding sites (Fig. 4B). When the entire core enhancer was deleted, virus production was reduced more than 1,000-fold (ΔNF-κB/Sp1, 0.07% ± 0.09% compared to 239wt). The amount of p27 obtained after transfection of COS-1 with the ΔNF-κB/Sp1 mutant was close to the detection limit of approximately 20 pg of p27 antigen per ml, and no significant virus production could be detected when additional upstream sequences were deleted (ΔNF-κB/Sp1/US364, 0.01% compared to 239wt [Fig. 4B]). Thus, with the same cell line and transfection protocol, the effect of deletions of the core enhancer region on virus production was more than 2 orders of magnitude higher than the effect on transcriptional activity in transient reporter assays (Fig. 3 and 4B). The deletion of 384 bp of the US region had little effect on luciferase reporter gene expression in the presence or absence of the core enhancer elements (Fig. 3). Nonetheless, this deletion resulted in a sixfold decrease in virus production in the context of an NF-κB-deleted provirus (ΔNF-κB, 45% ± 15%; ΔNF-κB/US384, 7.8% ± 1.7%). An 80-fold drop in p27 antigen levels was observed when the four Sp1 binding sites were removed (ΔSp1, 2.5% ± 1.8%; ΔSp1/US384, 0.03% ± 0.02% [Fig. 4B]).

Replication of SIVmac239 LTR mutants.

Virus stocks prepared from transient transfection of COS-1 cells were used to investigate the replicative properties of the SIVmac239 variants with deletions in the US region or mutations in the SF1 to SF3 factor binding sites. All mutants replicated with kinetics highly similar to those of the parental SIVmac239 clone in PHA-stimulated rPBMC, CEMx174 cells, MmHF7062A cells, and CB15 cells (Fig. 5 and data not shown). Thus, deletion of essentially all U3 sequences upstream of the NF-κB sites (400 bp) had no significant effect on replication of SIVmac. In the next set of experiments, we analyzed the replicative properties of SIVmac239 LTR mutants containing deletions in the US region in conjunction with deletions of the NF-κB or Sp1 binding elements. COS-1 transfection with proviral constructs missing the Sp1 binding sites resulted in inefficient virus production (Fig. 4B). Therefore, virus stocks generated from transfected CEMx174 cells were used for these experiments. PCR and sequence analysis at the end of culture confirmed the presence of the deletions (data not shown). All US deletion mutants replicated with efficiencies similar to that for the parental SIVmac239 clone in the four cell lines mentioned above when the single NF-κB binding site or the Sp1 binding sites were preserved (Fig. 6 and data not shown). The effective replication of the LTR variants containing US deletions in conjunction with the Sp1 deletion was striking, because these LTR variants showed up to 3,000-fold-lower amounts of p27 antigen production in transfected COS-1 cells (ΔSp1/US384, 0.03% ± 0.02% [Fig. 4B and 6]).

Next, we investigated if enhancer elements located in the US65 region have an effect on SIVmac replication when the entire core enhancer region is deleted. In agreement with previously published data (14), the NF-κB/Sp1 deletion had little effect on replication (ΔNF/Sp [Fig. 7]). Also, deletion of the 334 bp of upstream U3 sequences that were not preserved in vivo (22), in conjunction with the core enhancer region, did not reduce the replicative efficiency (ΔNF/Sp/NU [Fig. 7]). When the first 30 to 40 bp of the US65 region, including the SF1 site, were also removed, delayed replication kinetics were observed in rPBMC, CEMx174 cells, and MmHF7062A cells (ΔNF/Sp/US364 and ΔNF/Sp/US374 [Fig. 7]). Additional deletion of the next 10 bp (ΔNF/Sp/US384), affecting the SF2 and SF3 binding sites, further delayed and reduced virus production in rPBMC and MmHF7062A cells (Fig. 7A and C). In contrast, the ΔNF/Sp/US384 variant replicated with kinetics highly similar to those of the ΔNF/Sp/US364 and ΔNF/Sp/US374 variants in CEMx174 cells (Fig. 7B). The ΔNF/Sp/US400 variant, in which only 65 bp of the 517 bp U3 region (13%) are preserved, showed no detectable levels of reverse transcriptase activity in rPBMC (Fig. 7A). Low levels of replication, however, could be measured after infection of CEMx174 and MmHF7062A cells (Fig. 7B and C). Thus, SIVmac239 with only 95 of the 517 bp of the SIVmac U3 region (18%), containing only sequences required for integration, the TATA box, and a single NF-κB binding site, grows with normal kinetics in the four cell types analyzed (ΔSp1/US384 [Fig. 6 and data not shown]). Only the presence of the 26 bp usually located immediately upstream of the NF-κB site was required for efficient viral replication in the absence of the entire core enhancer (Fig. 7). The results show that enhancer elements located in the 26 bp upstream of the NF-κB site can compensate for the complete loss of the core enhancer.

Deletion of the SF2 and SF3 binding sites reduced viral replication in rPBMC and MmHF7062A cells (Fig. 7). Therefore, we tested if point mutations in these factor binding sites also affect replication. As shown in Fig. 8, mutations altering the SF2 (AGAAGG→TATGGG) or SF3 (GGAAA→ GG G AG) binding site resulted in delayed replication in CEMx174 and MmHF7062A cells compared to the ΔNF/Sp/NU control virus (Fig. 8). In these two cell lines, the variant containing mutations in all three SF sites replicated with an efficiency comparable to those of the ΔNF/Sp/mut-SF2 and ΔNF/Sp/mut-SF3 mutants. Mutations in the SF2 or SF3 site also reduced replication in rPBMC (Fig. 8A). The effects varied slightly between experiments performed with rPBMC from different animals. Nonetheless, these point mutations always caused a delay in viral replication, with the most drastic effect occurring when all three sites were altered.

The PuB2 site binds to Elf-1.

It has previously been demonstrated that two PuB sites located in the HIV-2 enhancer bind to Elf-1 and Ets-1 (24–26). SIVmac is closely related to HIV-2, and two purine-rich sequences can be identified at a similar location in the SIV enhancer (Fig. 2). One of these purine-rich regions, named SIV PuB1 in analogy to the HIV-2 LTR, is located just at the 3′ end of the NOD and was not conserved in the majority of animals infected with nef-defective SIVmac239 (reference 22 and Fig. 2). EMSAs with in vitro-translated Ets-1 and Elf-1 or cellular nuclear extracts were performed to test binding to SIV PuB1 and PuB2 (Fig. 9). Although the SIV PuB1 site is highly similar to the consensus binding site for members of the Ets family of transcription factors (Fig. 9E), it bound only weakly to Elf-1 (Fig. 9B, lane 7) and not to Ets-1. The presence or absence of the PuB1 site had no effect on viral replication or promoter activity (data not shown). The SIVmac PuB2 site completely overlaps the previously defined SF2 and SF3 binding sites (40) (Fig. 2 and 9E). It binds to Elf-1 (Fig. 9B, lane 10) but, in contrast to the HIV-2 PuB2 site, not to Ets-1 (Fig. 9A, lane 4, and data not shown). Binding to Elf-1 was reduced compared to the MSV and HIV- 2 PuB control probes (Fig. 9A, lane 5; Fig. 9B, lanes 4 and 10). Mutations in the conserved central GGA motif (SF3) abolished, and changes in the 5′ flanking region (SF2) of the SIV PuB2 site reduced, Elf-1 binding (Fig. 9B, lanes 13 and 16). Supershift assays with Elf-1 antibody confirmed that the PuB2 binding protein was indeed Elf-1 (Fig. 9B, lane 11). To demonstrate that Elf-1 binds also in the presence of other DNA binding factors, further gel retardation experiments were performed with nuclear extracts from MmHF7062A cells (Fig. 9C). As shown in Fig. 9C, lanes 2 and 11, incubation of nuclear extracts with probes corresponding to the PuB2 and the MSV probe resulted in similar band shifting patterns. While the lower band could also be observed with the PuB1 and mut-SF3 probe, the upper complex was specific for the MSV and PuB2 probes (Fig. 9C, lanes 2, 5, 8, and 11, upper complex indicated by asterisk). Addition of a monoclonal antibody which reacted specifically with Elf-1 (Fig. 9D) gave rise to a supershifted complex with the PuB2 and MSV probes but not with the PuB1 or mutSF3 oligonucleotide (Fig. 9C, lanes 3, 6, 9, and 12; supershift indicated by arrow). However, even in the presence of high amounts of Elf-1 antibody, the PuB2- and MSV-specific complex could not be disrupted entirely, indicating the binding of additional nuclear proteins to these sites (data not shown). We conclude from these data that Elf-1 binding to the PuB2 site can even be detected in the presence of a complex mixture of DNA binding proteins, but further, as yet unidentified factors contribute to the formation of nucleoprotein complexes on this sequence element.