A Naturally Occurring Substitution in Human Immunodeficiency Virus Tat Increases Expression of the Viral Genome

Abstract

A natural amino acid substitution in the human immunodeficiency virus type 1 (HIV-1) transcriptional activator Tat increases its activity and compensates for deleterious mutations elsewhere in the Tat protein. Substitution of asparagine for threonine 23 increases Tat transactivation of the HIV-1 promoter and the binding of Tat to the cellular kinase positive transcription elongation factor b (P-TEFb). Of nine other position 23 mutations tested, only the serine substitution retained wild-type activity. Correspondingly, asparagine is the most frequent amino acid at this position in HIV-1 isolates, followed by threonine and serine. Asparagine is prevalent in Tat proteins of viruses in clades A, C, and D, which are major etiologic agents of AIDS. We suggest that selection for asparagine in position 23 confers an advantage to the virus, since it can compensate for deleterious mutations in Tat. It may also support the replication of otherwise less fit drug-resistant viruses and permit the emergence of virulent strains.

Rapid viral human immunodeficiency virus type 1 (HIV-1) replication coupled with the high error rate of reverse transcription leads to a high mutation rate that is reflected in the genetic variability of the HIV genome (23). In an individual host, the swarm of related viruses, referred to as a “quasispecies,” is subject to selection by the host environment, and this can result in the emergence of new viral populations (15). The data presented here suggest that selection pressure is also exerted on the mechanism of viral transcription that depends on the HIV-1 regulatory protein Tat (transactivator of transcription).

Role of Tat in HIV infection.

Tat is essential for transcription of the HIV genome in activated T cells. After proviral integration, Tat interacts with the cellular protein complex P-TEFb (positive transcription elongation factor b) and the viral RNA element TAR (transactivation response), thereby recruiting the elongation factor to the transcription complex (8). P-TEFb, also known as TAK (Tat-associated kinase) (11), consists of the cyclin-dependent kinase CDK9 and cyclin T1 (21, 31). Cyclin T1 can bind directly to Tat and TAR (30). The complex phosphorylates the carboxy-terminal domain (CTD) of RNA polymerase II, leading to greatly increased production of full-length viral RNA (24).

The 72-amino-acid (aa) one-exon form of the Tat protein is functional in HIV transcription (6). Its first 48 aa comprise the activation domain and are sufficient for binding to P-TEFb (31). This domain is composed of acidic, cysteine-rich, and core regions (Fig. 1A). The basic region is necessary for binding to TAR, as well as for nuclear localization and bridging P-TEFb with its substrates (7, 13). At the C terminus of Tat72 is a glutamine-rich auxiliary domain that is nonessential for transcriptional activation (10).

FIG. 1.

Transactivation by Tat mutants. (A) Tat domains and mutants. The wild-type (WT) NL4-3 Tat sequence is shown (top line) with its domains. Tat-WHA is a naturally derived Tat variant (16) differing from wild-type Tat at the eight sites marked. Tat 39-42, T23N, T23N-3942, 5963, A67V and 5963A67V contain subsets of the WHA changes in a wild-type Tat background. Mutations were created by site-directed mutagenesis (26). (B) Increasing amounts of plasmid expressing wild-type (WT) Tat or Tat T23N were transfected into 293 cells in 12-well plates (1.5 × 10⁵ cells per well) together with 100 ng of LTR-luciferase and 300 ng of RSV-Renilla plasmids (26). The ratio of firefly to Renilla luciferase activity is plotted as relative luciferase units (RLU). Values are averages of two measurements with standard errors. (C) Plasmid expressing wild-type or mutant Tat (400 ng) was cotransfected with 200 ng of LTR-luciferase and 600 ng of RSV-Renilla reporters into 293 cells (4 × 10⁵). The LTR activity of each sample is shown at the top of the chart as a percentage of wild-type activity. Results are the average of two experiments.

Natural variations in Tat sequence.

Sequence analysis of Tat variants from peripheral blood mononuclear cells (PBMCs) reveals rapid HIV-1 evolution in the host as well as ex vivo (16). Cells harboring proviruses with completely defective Tat mutants do not produce progeny viruses (18, 28), but cells containing proviruses encoding attenuated Tat proteins may allow viral replication if the cells are activated and their transcription machinery is operating optimally (26). A small number of infected T cells that escape apoptosis gradually revert to a resting state and become memory cells (22). During this process, attenuated Tat proteins may not support viral transcription, and the provirus will persist latently in the infected memory cell until its full reactivation.

Conversely, mutations in Tat that improve its activity may arise in the same fashion and could increase the efficiency of viral replication. A fast-replicating, highly cytopathic virus isolated from an AIDS patient's PBMCs carried an amino acid change in the Tat basic domain that increased its activity in transactivation and viral replication assays (5, 14), although the biochemical basis for this effect was not established. It is possible that in immunocompromised patients, such virulent viruses could replicate faster and cause more rapid depletion of CD4⁺ T cells in the host, leading to accelerated disease progression. Transmission of a virulent variant to a new host might then accelerate progression to AIDS in the recipient (27).

We recently found that the natural Tat variant WHA (16) is almost as active as wild-type Tat NL4-3 (26). WHA Tat differs from the wild-type protein at eight sites (Fig. 1A), including three in the core domain that debilitate Tat function when introduced as a group into the wild-type protein (Tat 39-42). This observation implies that other changes in WHA, located in the cysteine-rich and auxiliary domains, increase Tat activity and compensate for the deleterious mutations in Tat 39-42.

Asparagine 23 increases Tat transactivation.

Whereas mutations in the auxiliary domain generally have little or no effect on transactivation (10), the cysteine-rich domain is part of the Tat activation domain, which is more highly conserved and binds P-TEFb. Reasoning that the single change in the essential cysteine-rich domain of WHA was more likely to have a large influence on Tat activity than changes in the auxiliary domain, we substituted asparagine in place of threonine at position 23 in the wild-type vector. The resultant mutant, Tat T23N, was compared to wild-type Tat in transactivation assays by monitoring the expression of luciferase driven by the HIV long terminal repeat (LTR) in human 293 cells. Over a broad range of concentrations, Tat T23N was approximately threefold more effective than wild-type Tat in stimulating expression from the HIV-1 promoter (Fig. 1B). We also constructed three auxiliary domain variants to examine the contribution of the other changes in WHA (Fig. 1A). Three of these changes (residues 61, 63, and 67) are poorly conserved in natural HIV isolates, and the changes are to common variations (26). As expected, they had only a minor effect on transactivation (Fig. 1C). Tat 5963A67V, which contains all four of the auxiliary domain changes present in WHA, reduced the activity of wild-type Tat by one-third, while Tat 5963 and Tat A67V were less inhibitory.

These results indicate that the T23N substitution is solely responsible for up-regulating the activity of Tat WHA. To determine whether it can compensate for the defect resulting from the core domain changes in WHA Tat, we introduced it into Tat39-42, generating Tat T23N-3942 (Fig. 1A). Tat T23N-3942 elicited nearly wild-type levels of expression from the HIV LTR (Fig. 1C), showing that asparagine 23 can rescue the attenuated activity of Tat 39-42.

To determine whether this substitution functions to elevate expression from the LTR in the context of the molecular clone, we replaced threonine 23 in the Tat gene of the NL43-LucE⁻ molecular clone with asparagine and produced pseudotyped viruses coated with the murine leukemia virus (MuLV) amphoteric envelope protein (4). In the NL43-LucE⁻ clone, part of the Nef gene is replaced by the firefly luciferase gene, and the envelope gene is mutated. U937 and CEM cells were infected with these viruses, and luciferase activity was measured at different time points after infection (Fig. 2). The virus carrying the Tat T23N mutation was approximately fivefold more effective at inducing transcription than virus carrying wild-type Tat over a range of multiplicities. Hence, the Tat T23N substitution is more potent when it is expressed from molecular clone DNA after infection.

FIG. 2.

Viral gene expression. U937 cells (A) or CEM cells (B) were infected with the indicated amounts of pseudotyped HIV-1 virus carrying either wild-type Tat or Tat T23N (4). Pseudotyped viruses were produced in 293T cells by transfection with pNL4-3-LucE⁻ (wild type or T23N) and pSVL-MEA (encoding the MuLV envelope protein) and were quantified by viral p24 assay (Zeptomatrix, Buffalo, N.Y.). Cells were harvested at the indicated time points (hours post infection [HPI]), and extracts were assayed for luciferase activity and total protein. Luciferase activity (arbitrary units [AU]) is plotted relative to the total protein in each sample.

Tat position 23 affects P-TEFb binding.

The location of residue 23 in the activation domain of Tat suggested that the increased transactivation activity displayed by Tat T23N could be explained by an increased ability to bind P-TEFb. Figure 3A compares the amounts of P-TEFb pulled down from 293 cell extracts by equivalent amounts of glutathione S-transferase (GST) fusions containing wild-type or mutant Tat proteins. Bound P-TEFb was determined by Western blotting for CDK9 and by TAK assays for CTD kinase activity. In other experiments, we showed that cyclin T1 accompanies CDK9 and that P-TEFb binding is required for phosphorylation of the CTD substrate (25, 26). GST-Tat T23N bound about threefold more CDK9 than wild-type GST-Tat and accordingly gave approximately threefold more phosphorylation of the CTD4 peptide. Tat 39-42 bound P-TEFb less well than wild-type Tat, as shown previously (26), and this deficiency was restored in Tat T23N-3942. Similar results were obtained with U937 cell extracts (Fig. 3B). Thus, the introduction of asparagine in position 23 increased P-TEFb binding in the wild-type Tat context as well as in the presence of core domain mutations that reduce P-TEFb binding and attenuate Tat function.

FIG. 3.

Effect of T23N substitution on P-TEFb binding. (A) Equal amounts of wild-type (WT) and mutant GST-Tat proteins were immobilized on glutathione-Sepharose beads. The top panel shows the immobilized proteins separated in a sodium dodecyl sulfate-polyacrylamide gel and visualized with Coomassie brilliant blue stain. The GST-Tat beads were incubated with ∼40 μg of 293 whole-cell extract (26), and the bound protein was analyzed by immunoblotting with anti-CDK9 antibody (middle panel) or assayed for P-TEFb kinase (TAK) activity with CTD4 substrate (bottom panel) (25, 26). The bar graph shows quantitation of kinase activity. Panel B is the same as panel A, but with 400 μg of phorbol myristate acetate-stimulated U937 cell extract.

In natural viral isolates, the cysteine-rich domain of Tat contains several invariant residues and other positions where alterations are tolerated (26). Since position 23 is in the variable category, we examined the effect of alternative substitutions at this site. Like asparagine, the amino acids alanine, isoleucine, proline, and serine can result from a single nucleotide change in the threonine codon ACC. Replacement of threonine with alanine, isoleucine, or proline reduced Tat transactivation to ∼30 to 40% of the wild-type level (Fig. 4A). Correspondingly, these three mutant Tat proteins were defective for P-TEFb binding and CTD4 phosphorylation in pull-down assays from U937 or 293 cell extracts (Fig. 4B). In contrast, the T23S substitution mutant was approximately as active as wild-type Tat in all assays. These data indicate that position 23 is involved in binding P-TEFb and that the uncharged polar side chains of threonine, serine, and asparagine are strongly preferred for this interaction. The side chain requirement was further explored by examining the effects of several residues that resemble asparagine in size, polarity, or chemical reactivity. As seen in Fig. 4C, the substitution of glutamine, aspartate, glutamate, or valine for threonine at position 23 reduced Tat transactivation activity to ∼30 to 40% of that of the wild type. The activity of Tat T23H was even lower at ∼20%. Surprisingly, the polar residues had nearly the same effect as the hydrophobic residues, and even closely related amino acids were much less active than asparagine at this position.

FIG. 4.

Effects of position 23 substitutions on Tat transactivation and interaction with P-TEFb. (A) Transactivation by the indicated Tat mutants was assayed as in Fig. 1C (using 100 ng of Tat vectors) and expressed as a percentage of wild-type (WT) activity. The activity of each mutant relative to wild-type Tat is given above the chart. (B) P-TEFb binding to GST-Tat mutants was monitored by immunoblotting with anti-CDK9 antibody, and TAK activity was assayed by CTD4 phosphorylation as in Fig. 3. (C) Transactivation by an additional series of Tat mutants was assayed as in panel A.

These results indicate a high degree of specificity in the interaction between P-TEFb and the residue at position 23. The Tat-cyclin T1 interaction is postulated to require five cysteine residues and a histidine in the Tat cysteine-rich domain as well as Cys-261 in the Tat-TAR binding region of cyclin T1 (9, 12). In one model, cysteine residues of the two proteins cooperate in coordinating zinc. Position 23 lies within the highly flexible (1) cysteine-rich region of Tat, adjacent to an essential cysteine residue, and both are needed for Tat activity (3). Our data imply that an amino acid with a small hydrophobic side chain is important at this position, either for stabilizing the structure of Tat or its complex with cyclin T1. Increasing the binding ability in one region can compensate for the decrease in another (Fig. 3A). Whether this implies that the two regions combine to form a single binding site or that two-point Tat-cyclin T1 interactions occur awaits more detailed analysis.

Biological significance of Tat sequence variation.

The diversity of Tat sequences is reflected in the consensus sequences for the five clades shown in Fig. 5A. While Tat is highly variable, certain residues are conserved among all clades, especially in the cysteine-rich core and basic domains. As pointed out previously (26), when conservation is less than absolute, usually only a restricted number of alternative amino acids occur at a significant frequency. Natural HIV-1 isolates contain asparagine (53%), threonine (36%), or, less commonly, serine (11%) in position 23 (26). Interestingly, asparagine is commonly found in this region of HIV-2 and other lentiviral Tat proteins (Fig. 5B). In HIV-1, the codons used are AAC and AAU for Asn, ACC for Thr, and AGC for Ser. Of these, the most frequent codon in the sequence set examined is AAC, and all of the others can be reached by a single change from this one. Consistent with selection for functional Tat, the most frequent residue at this position (Asn) is the most active. The other two amino acids found naturally at this position (Thr and Ser) are more active than the eight alternatives tested. While the spectrum of substitutions tested here is not exhaustive, the absence of other residues at this position in 135 full-length viral sequences implies that only the three amino acids retained by selection are compatible with Tat function. It is particularly notable that glutamine, aspartate, and valine are neither fully active nor evolutionarily selected despite their similarity to asparagine. Evidently the differences in side chain length or chemistry are sufficient to perturb the structure of Tat such that it cannot adequately fulfill its function.

FIG. 5.

Tat sequences from different HIV-1 clades and related lentiviruses. (A) Clade consensus sequences. The amino acid sequence of Tat (aa 1 to 48) from different HIV-1 subtypes is compared with that of the wild-type Tat (NL4-3) used in this study. The consensus sequence of each clade was obtained from the Los Alamos HIV Sequence Database (http://hiv-web.lanl.gov/). The number of unique sequences used to derive each consensus is indicated in parentheses. Amino acids that differ from Tat NL4-3 are marked, and residue 23 is shaded. Clade G sequence is ambiguous for both asparagine and lysine at position 7. (B) Beginning of the cysteine-rich region in primate lentiviruses. Conserved residues are boxed. A dash signifies missing residues, and x represents an unspecified nonconserved residue. Position 23 of HIV-1 Tat lies between the first two highly conserved cysteine residues, C22 and C25, in the cysteine-rich domain. Both positions 23 and 24 are frequently asparagine in HIV-1 Tat (53 and 42% of the sequences in our analysis [shown as n]) as well as in related chimpanzee viruses. In HIV-2 and viruses from several other primate and simian species, three amino acids separate the corresponding cysteines, and the middle residue is invariably N.

Our findings identify residue 23 in the cysteine-rich domain of Tat as a position that can confer replicative advantage upon HIV-1 as a result of a single nucleotide change. The presence of Asn-23 may be associated with two distinct phenomena: increasing fitness or virulence and counterbalancing internal Tat mutations that on their own tend to diminish Tat activity. Precedents support both of these mechanisms. In addition, it is conceivable that Tat variants with greater activity may be selected for in viruses that harbor compromised reverse transcriptase and/or protease as a result of exposure to antiretroviral therapy (2). Intrahost fluctuations in the sequence of Tat can result in the emergence of strains with greater Tat activity that might lead to more rapid progression to AIDS. A mutation in the Tat basic domain that emerged during disease progression was shown to increase the HIV replication rate in culture (14), and Tat proteins from virulent African strains were found to have elevated transactivation activity (20). Asparagine 23 has become fixed in clades A, C, D, and G, which predominate in Africa (Fig. 5A). During forced evolution in culture, a second-site mutation was generated in Tat that partially restored activity to a severely defective Tat mutant (29). In some naturally occurring sequences, like WHA examined here, the N23 residue provides second-site compensation for deleterious mutations in the Tat core region. Thus, asparagine 23 may act as a buffer, allowing for sequence diversity in the core region that confers resistance to the immune system (19). The observation that immunization against a simian/human hybrid virus (SHIV) did not efficiently protect against infection by a more virulent SHIV strain (17) underlines the importance of understanding parameters that influence viral pathogenesis.