Dysregulation through the NF-κB Enhancer and TATA Box of the Human Immunodeficiency Virus Type 1 Subtype E Promoter

Abstract

The global diversity of human immunodeficiency virus type 1 (HIV-1) genotypes, termed subtypes A to J, is considerable and growing. However, relatively few studies have provided evidence for an associated phenotypic divergence. Recently, we demonstrated subtype-specific functional differences within the long terminal repeat (LTR) region of expanding subtypes (M. A. Montano, V. A. Novitsky, J. T. Blackard, N. L. Cho, D. A. Katzenstein, and M. Essex, J. Virol. 71:8657–8665, 1997). Notably, all HIV-1E isolates were observed to contain a defective upstream NF-κB site and a unique TATA-TAR region. In this study, we demonstrate that tumor necrosis factor alpha (TNF-α) stimulation of the HIV-1E LTR was also impaired, consistent with a defective upstream NF-κB site. Furthermore, repair of the upstream NF-κB site within HIV-1E partially restored TNF-α responsiveness. We also show, in gel shift assays, that oligonucleotides spanning the HIV-1E TATA box displayed a reduced efficiency in the assembly of the TBP-TFIIB-TATA complex, relative to an HIV-1B TATA oligonucleotide. In transfection assays, the HIV-1E TATA, when changed to the canonical HIV-1B TATA sequence (ATAAAA→ATATAA) unexpectedly reduces both heterologous HIV-1B Tat and cognate HIV-1E Tat activation of an HIV-1E LTR-driven reporter gene. However, Tat activation, irrespective of subtype, could be rescued by introducing a cognate HIV-1B TAR. Collectively, these observations suggest that the expanding HIV-1E genotype has likely evolved an alternative promoter configuration with altered NF-κB and TATA regulatory signals in contradistinction with HIV-1B.

Human immunodeficiency virus type 1 (HIV-1) subtype B was the virus initially described in countries such as India, Thailand, and the Republic of South Africa; however, the current heterosexual epidemics in those countries are caused by other HIV-1 genotypes that entered later (8, 23, 25, 27, 28). Virtually all new heterosexually transmitted HIV infections in Thailand are now HIV-1E, and among intravenous drug users the relative proportion of HIV-1E has been reported to be increasing (24), indicating that HIV-1E has competed more efficiently than the HIV-1B genotype in that setting (14).

Regulated transcription of HIV-1 is essential to the establishment of a productive infection. HIV-1 expression can be dramatically influenced by apparently subtle nucleotide changes within the promoter region, which includes the TATA box, an essential DNA element necessary for recruitment of TATA binding protein (TBP) and initiation of RNA synthesis; the NF-κB enhancer, a tandem DNA binding site recognized by the positive host cell regulator NF-κB:p50:p65; and the RNA enhancer TAR, to which the viral transactivator Tat binds (for a review, see reference 7). In addition to their roles in recruiting unique factors, these sites juxtapose nucleic acid binding proteins that participate in protein-protein interactions, for example, TAT-TBP (11) and Rel-TBP (12).

The HIV-1E genotype contains a distinct regulatory architecture, suggesting potentially important differences in viral regulation (16). Notably, NF-κB:p65 (RelA)-dependent activation of HIV-1 transcription was shown to be correlated with the copy number of the NF-κB enhancer, such that subtype E isolates which contain one κB site were consistently less inducible than subtype B isolates that contain a standard two NF-κB sites. The copy number of the NF-κB enhancer is likely to influence replication rate, since viruses which contain two tandem NF-κB sites replicate with higher efficiency than κB mutant viruses (4).

Since the HIV-1E subtype is spreading efficiently, the presence of a single NF-κB site within the HIV-1E promoter prompted us to determine whether physiologically relevant activators, such as tumor necrosis factor alpha (TNF-α), might nevertheless efficiently activate HIV-1E. Many studies have implicated an important, if not central, role for the immunomodulatory cytokine TNF-α both in the activation of HIV-1 gene expression and associated pathogenic sequelae of HIV-1 infection. TNF-α-mediated activation of HIV-1 has been linked to the induction of Rel heterodimer p50:p65 nuclear translocation and to subsequent binding activity at the NF-κB enhancer (2, 18).

An additional, peculiar feature of the HIV-1 subtype E promoter is the prevalence of both a variant TATA box (ATAAAA), in contrast with the more common TATA box (ATATAA), and a variant TAR bulge-loop region that contains a nucleotide deletion flanked by two polymorphisms. Previous studies designed to assess the role of the TATA box within the context of the HIV-1B subtype evaluated mutants that resemble the subtype E TATA (E-TATA) sequence and were shown to dramatically reduce transcriptional activity (3). The prevalence of this naturally occurring HIV-1E TATA sequence variant would therefore seem to imply potentially reduced activity.

Stable and distinct NF-κB enhancer and TATA-TAR configuration among HIV-1E primary isolates.

To confirm whether previously observed differences in the HIV-1E promoter are stable, we sequenced an additional 10 epidemiologically unrelated isolates. All HIV-1E isolates contained a defective NF-κB II site, as previously observed (Fig. 1a). In addition, substitutions originally noted in the TATA box and the TAR region were also confirmed, such that 14 of the 15 HIV-1E isolates contained the HIV-1E-specific TATA box (ATAAAA) as well as substitutions in the TAR bulge-loop region. To test the comparative induction of these promoter sequences, lacZ reporter genes were created (Fig. 1b) that contain naturally occurring long terminal repeat (LTR) sequences or LTR sequences with replacements in the following regions: the TATA box of HIV-1E (E18ltr.t), the TAR bulge-loop region (E18ltr.tb and E18ltr.b), and the NF-κB II site (E18ltr.κB). To test the role of Tat activation, the first exon (exon 1) of primary HIV-1E and HIV-1B isolates was PCR isolated and cloned into an expression vector, as indicated (Fig. 1b).

FIG. 1.

DNA sequence alignments of the κB enhancer region through the TATA-TAR region (−108 to +46) and plasmid constructions used in this study. (a) The region encompassing the NF-κB sites among HIV-1B and HIV-1E isolates is shown. Each NF-κB site is shaded. Note that all HIV-1E isolates contain a defective NF-κB II site. The region encompassing the TATA box through TAR is also shown. All HIV-1E isolates and the HIV-1A reference isolate, U455 (17), contain a nucleotide deletion (T25Δ) predicted to yield a 2-nucleotide bulge relative to HIV-1B. All HIV-1E isolates also contained two additional substitutions within the TAR bulge region, A22G and T31C. (b) Reporter gene constructs containing primary and mutated LTR sequences were made by oligonucleotide-directed mutagenesis and cloned directly into the KpnI-HindIII site of the pBgal-Basic vector (Clontech, Palo Alto, Calif.), as indicated. HIV-1E Tat and HIV-1B Tat exon 1 sequences were engineered into the HindIII-PstI site of the expression vector pCDNA3.1/Zeo (Invitrogen, Carlsbad, Calif.). Peripheral blood mononuclear cell-derived DNA samples from heterosexually transmitted isolates J35 to J48 were obtained from R. Sutthuent (Mahidol University, Bangkok, Thailand). Samples were sequenced by ABI automated sequencing, as previously described (GenBank accession no. AF080159 through AF080168). The sequences of the internal DNA oligonucleotides used to create the mutant sites and Tat plasmids are as follows: E18-Btata+, 5′-CAG ATG CTG CAT ATA AGC AGC CGC T-3′ and E18-Btata−, 5′-AGC GGC TGC TTA TAT GCA GCA TCT G-3′; E18-Bbulge+, 5′-GAC CAG ATC TGA GCC TGG GAG CTC T-3′; E18-Bbulge−, 5′-AGA GCT CCC AGG CTC AGA TCT GGT C-3′; E18BkB+, 5′-TTC TAC AAG GGA CTT TCC GCT GGG GAC-3′; E18BkB−, 5′-GTC CCC AGC GGA AAG TCC CTT GTA GAA-3′. HindIII-Etat+, 5′-CCA AGC TTA CCT GCC ATG GAG CCG GTA GAT CCT AAC CTA GAG CCC-3′; PstI-Etat−, 5′-A AAC TGC AGT TAC TGC TCT GGT ATA GGA TTT TGA TGA TCC-3′; HindIII-Btat+, 5′-CCA AGC TTA CCT GCC ATG GAG CCA GTA GAT CCT AGA CTA GAG CCC-3′; PstI-Btat−, 5′-A AAC TGC AGT TAC TGC TTT GAT AAA AAA ACT TGA TGA GTC-3′.

HIV-1E displays reduced TNF-α cytokine responsiveness in correlation with a defective upstream NF-κB II site.

Because the NF-κB enhancer copy number differs between the B and E subtypes and since TNF-α cytokine activation has been shown to be reliant upon the NF-κB enhancer, we assessed the TNF-α-mediated induction of representative subtype promoters. As shown in Fig. 2, the HIV-1E subtype, which contains one functional NF-κB site, displayed reduced TNF-α responsiveness relative to HIV-1B in both Jurkat T cells (Fig. 2a; compare lanes 1 and 2 with 3 and 4) and 293 cells (Fig. 2b; compare lanes 1 to 3 with 4 to 6).

FIG. 2.

Variant NF-κB and TATA sequences influence activation of the HIV-1E LTR. The HIV-1E subtype displayed reduced TNF-α responsiveness, relative to HIV-1B, in both Jurkat T cells (panel 2a; compare lanes 1 and 2 with 3 and 4) and 293 cells (panel b; compare lanes 1 to 3 with 4 to 6) in cotransfection assays. Replacement of the defective upstream site improved TNF-α response (panel b; compare lanes 7 to 9 with 4 to 6). Activation of an HIV-1E chimeric promoter containing the B-TATA (E18ltr.t) by both subtype Tat’s was unexpectedly reduced in 293 cells compared to that of the wild-type HIV-1E LTR construct (E18ltr) (panel c; compare lanes 11 to 15 with 6 to 10). Addition of a compensatory B-TAR (E18ltr.tb) restored Tat-mediated activation (panel c, lanes 16 to 20). Replacement of the HIV-1E TAR with HIV-1B TAR alone did not appreciably influence Tat activation (panel c, lanes 21 to 25). Fluorescence units and fold activations are indicated and represent an average of duplicates from representative transfections. Target plasmids were transfected with 100 ng of genomic DNA/10⁵ 293 cells or 10 μg of genomic DNA/5 × 10⁶ Jurkat T cells. Expression vectors were transfected in the amounts indicated. TNF-α (Genzyme, Cambridge, Mass.) stimulation was performed 18 to 24 h posttransfection at concentrations of 5 to 100 ng/ml in both Jurkat and 293 cells. Cells were assayed 48 h posttransfection for β-galactosidase activity by using the methylumbelliferyl-β-glucuronide assay. Fluorescence was determined with a Fluoroskan plate reader (Flow Laboratories, Helsinki, Finland). •, 2-nucleotide TAR bulge.

Repair of the defective NF-κB II site in the HIV-1E LTR improves TNF-α-mediated induction.

We speculated that the absence of an upstream NF-κB II site within HIV-1E might account for the reduced TNF-α response. As shown in Fig. 2b, replacement of the defective upstream site improved TNF-α response (compare lanes 7 to 9 with 4 to 6), thereby suggesting a direct role for NF-κB enhancer copy number and TNF-α-dependent transcriptional activation.

A single nucleotide substitution of the E-TATA box, yielding a canonical B-TATA sequence (ATAAAA→ATATAA), unexpectedly reduces HIV-1 Tat induction independent of Tat subtype and can be rescued with compensatory changes within TAR.

The HIV-1E TATA box contains a single stable nucleotide polymorphism which distinguishes it from other HIV-1 subtypes (ATAAAA versus ATATAA [difference underlined]). Although absent in other HIV-1 subtypes, this variant TATA sequence is present in simian immunodeficiency virus (SIV) strains among African Green monkeys (9). The HIV-1E TAR sequences contain three associated nucleotide changes: a 2-nucleotide bulge (U25Δ) and two flanking variant nucleotides (A22G and U31C [see molecular model of TAR in Fig. 3b]) predicted to be in close proximity with bound Tat protein (13, 26). As has been previously noted, the HIV-1A subtype and certain SIV isolates also contain a 2-nucleotide bulge (6). To test whether the altered TATA and/or TAR sequences in HIV-1E represent nonneutral genetic substitutions, we created chimeric LTR-driven reporter genes which replaced the E-TATA with a “B-TATA” (E18ltr.t) and a “B-TAR” (E18ltr.tb) within the context of the HIV-1E LTR. Since HIV-1 Tat function has been previously shown to be sensitive to both TATA and TAR sequences, we chose to assess both HIV-1B Tat and a cognate HIV-1E Tat for activation of these constructs. As shown in Fig. 2c, activation of the HIV-1E chimeric promoter containing the B-TATA (E18ltr.t) by both subtype Tat’s was unexpectedly reduced compared to that of the wild-type HIV-1E LTR construct (E18ltr) (Fig. 2c; compare lanes 11 to 15 with 6 to 10). This may suggest that the HIV-1E TATA sequence represents a context-dependent adaptation necessary for optimal Tat function. Since Tat protein has been shown to interact with TBP, a component of TFIID (11), TBP-Tat activity might be influenced by the nucleotide sequence and genetic context of the TATA box and TAR. We reasoned, therefore, that an altered TATA sequence may require compatible TAR changes for efficient TBP-Tat complex function and activity. As shown in Fig. 2c (lanes 16 to 20), the presence of a compensatory B-TAR bulge-loop-containing reporter gene (E18ltr.tb) restored Tat-mediated activation. Interestingly, replacement of the HIV-1E TAR with HIV-1B TAR alone did not appreciably influence Tat activation (Fig. 2c, lanes 21 to 25). This suggests that the activity of TATA-TBP and Tat-TAR complexes is guided by genetic context.

FIG. 3.

Gel shift analysis of TBP-TFIIB-TATA assembly (a), TAR secondary structure (b), and phylogenetic divergence (c). (a) In gel shift assays, assembly of the TBP-TFIIB-TATA complex on the B-TATA oligonucleotides was dose responsive with increasing levels of TFIIB recombinant protein (left panel, 40 ng of TBP plus 0, 15, and 30 ng of TFIIB), while minimal assembly occurred on the E-TATA oligonucleotides in the presence of increasing levels of TFIIB protein (right panel, 40 ng of TBP plus 0, 15, and 30 ng of TFIIB). Recombinant TATA-binding protein and TFIIB (Santa Cruz Biotechnology) were incubated with radiolabelled HIV-1B or HIV-1E TATA oligonucleotides (24-mers, −39 to −15 relative to transcription start) for 30 min at room temperature in a total of 25 μl in binding buffer (50 mM Tris-HCl, 10 mM MgCl₂, 100 mM NaCl, 1 mM dithiothreitol, 100 ng of bovine serum albumin per μl, 0.01% Nonidet P-40, and 300 μg of dG-dC). DNA-protein complexes were resolved in 5% native polyacrylamide gels for approximately 1 h at 200 V, dried, and exposed to film. (b) Predicted secondary structure of the TAR bulge-loop region for HIV-1B (B-TAR), HIV-1A (A-TAR), and HIV-1E (E-TAR). The HIV-1B subtype contains a 3-nucleotide bulge, whereas the HIV-1A and E subtypes each contain a 2-nucleotide bulge. E-TAR differs from both A-TAR and B-TAR in having two additional nucleotide substitutions, A22G and U31C (arrows indicate differences from HIV-1B TAR). Also shown is a model of the HIV-1B TAR structure to denote positions of variant nucleotides (generated with InsightII software [MSI, San Diego, Calif.]). (c) Minimum evolution phylogenetic analysis (γ = 0.5 [F84 model]) of selected sequences using either the complete LTR (−450 to +46, entire U3 through TAR [left]) or the TATA-TAR region (−70 to +46 [right]), as indicated. Note that the entire LTR resolves distinct subtypes, whereas the TATA-TAR core places the HIV-1E subtype as an outlier from the other subtypes (note bootstrap values).

E-TATA and B-TATA oligonucleotides differ in assembly of the TBP-TFIIB-TATA complex.

To further investigate a potential role for the variant HIV-1E TATA box, we chose to determine, in gel shift assays, whether early steps in RNA polymerase II (Pol II) recruitment were influenced by assessing the capacity for recombinant TBP-TFIIB assembly to occur on B-TATA and E-TATA oligonucleotides. Many studies have previously shown that TFIIB plays a critical role in the assembly of the Pol II holocomplex by serving as a bridge between TBP and Pol II (22). As shown in Fig. 3a, assembly of the TBP-TFIIB-TATA complex on the B-TATA oligonucleotides was dose responsive with increasing TFIIB concentration, while a minimal effect on assembly occurred on the E-TATA oligonucleotides. Altered assembly may suggest that the distinct subtype TATA boxes differ in preinitiation complex formation and possibly recruitment of TBP-associated factors.

Comparison of the TATA-TAR regions indicates that subtype E contains unique genetic features and has diverged from the other subtypes, including subtype A.

Both subtype A and E TAR sequences have been described as containing a 2-nucleotide bulge (U25Δ) based on RNA folding criteria, as opposed to the 3-nucleotide bulge expected with the HIV-1B TAR (6). The A and E subtype similarity within the bulge region is provocative, since analysis of the entire genome of the E subtype reveals an A-E recombinant structure (6), with env and LTR regions being distinct from those of subtype A and potentially coselected. The possibility that the bulge region of the E subtype might be functionally linked with the TATA box polymorphism prompted a closer analysis of subtype A, which has a “B-like” TATA and an “E-like” 2-nucleotide bulge. The secondary structure prediction of the bulge-loop region (Fig. 3b) reveals that the HIV-1A bulge region differs from the B subtype solely in having a 2-nucleotide bulge, while the E subtype differs from both the B and A subtypes by containing two additional substitutions (A22G and U31C). A comparative phylogenetic analysis of the entire LTR region with the TATA-TAR region (Fig. 3c; compare left and right phylograms) supports the notion that the HIV-1E TATA-TAR is distinct while all other subtypes collapsed into a monophyletic group (note bootstrap values). This may suggest that the HIV-1E TATA-TAR region has undergone distinct genetic changes that may have been required for optimal Tat activity.

Divergent viral genotypes of HIV have clearly played an important role in both transmission efficiency and natural history. Studies conducted by our group have previously established that HIV-1 was transmitted 5- to 10-fold more efficiently than HIV-2 in the same cohort of female sex workers (10). Similarly, studies by others have shown that mother-to-infant transmission was much less frequent with HIV-2 (1, 5). We also have observed differences in virulence between HIV-1 and HIV-2 (15) and recently among different subtypes of HIV-1 (9a).

This study focuses on specific features of HIV-1E and extends previous observations by our group that described functional and architectural distinctions within the promoter sequences of expanding subtypes. We observed distinctions in the NF-κB enhancer copy number that appeared to confer a differential and correlated response to the inflammatory cytokine TNF-α, a potent and critical activator of HIV-1 gene expression. Genetic changes within the TATA and TAR regions also appear to have undergone context-adaptive changes to potentially maintain Tat function. Collectively, these observations support the notion that genetic divergence between the subtypes can provide a capacity for altered transcriptional activation and preinitiation complex assembly.

The dysregulation of TNF-α response by HIV-1E was correlated with a defective upstream NF-κB II site, since a mutant that restores this upstream site improved TNF-α response. This phenotype may suggest that the HIV-1E genotype, which appears to be quite efficient at spreading throughout southeast Asia, may have undergone genetic changes allowing for an alternative transcriptional strategy by differentially utilizing known, as well as potentially unidentified, transcriptional control mechanisms. Evidence for differential regulation, particularly gain-of-function transcription, may help to elucidate a causal link between transcription strength, viral replication, and ultimately epidemic spread. Recently, the GLI-2/THP-1 transcription factor has been demonstrated to augment activation of the HIV-1 LTR by Tat (3a). We have observed a differential gain of function with HIV-1E (and HIV-1C) LTR targets relative to HIV-1B response in transfection studies (unpublished data). Such novel gain-of-function mechanisms of transcriptional control may play functional roles in the apparent differential spread observed among the subtypes expanding globally.

Perhaps 10% of the HIV-1 isolates identified to date represent intergenotype recombinants (19–21). The HIV-1E subtype, for example, contains unique envelope and LTR sequences, while the rest of the viral genome represents subtype A sequence. Recombinant genomes may introduce novel genetic configurations that impact viral function. While this study focuses on genetic configurations that influence activity within the LTR, the ever-increasing number of intergenotypic recombinants being identified in other loci raises a larger issue regarding what role altered genetic contexts may play in the pathogenetic evolution of HIV-1.

A remaining question concerns whether a single introduction of HIV-1 occurred from SIVs resident among nonhuman primates in Africa or whether multiple introductions have occurred (the greatest likelihood is that there is no single common ancestor for all subtypes but that some subtypes, e.g., B and D, have a common progenitor). Provocatively, recent phylogenetic analysis of an HIV-1 sequence from a 1959 plasma sample (Z59) in Kinshasa place this isolate near the ancestral node of HIV-1B, -D, and -F and have prompted the conjecture that this sequence might represent the founding or a closely related founding viral genotype in humans (29). If Z59 represents a founding genotype, then subtypes such as HIV-1E (and HIV-1C), which are currently overtaking HIV-1B, may represent more recent promoter configurations that are potentially adapted for more efficient spread within the human population.