|Home | About | Journals | Submit | Contact Us | Français|
Elite suppressors (ES) are human immunodeficiency virus type 1 (HIV-1)-infected patients who maintain viral loads of <50 copies/ml without treatment. The observation that the HLA-B*57 allele is overrepresented in these patients implies that HIV-1-specific CD8+ T cells play a key role in suppressing viral replication. We have previously shown that while CD8+ T-cell escape mutations are rarely seen in proviral Gag sequences in resting CD4+ T cells from peripheral blood, they are present in every clone amplified from the low levels of free virus in the plasma of HLA-B*57+ ES. In this study, we compared the pattern of mutations in Nef sequences amplified from peripheral blood CD4+ T cells and from plasma virus. We show that Nef mutations are present in plasma virus but are rare in the cellular sequences and provide evidence that these plasma Nef variants represent novel escape mutants. The results provide further evidence of CD8+ T-cell-mediated selective pressure on plasma virus in ES and suggest that there must be ongoing HIV-1 replication in spite of the very low viral loads seen for these patients.
The mechanisms involved in the control of human immunodeficiency virus type 1 (HIV-1) replication in elite suppressors (ES) have yet to be fully defined. Certain major histocompatibility complex class I alleles such as HLA-B*27 and HLA-B*57 are overrepresented in ES (12, 25, 32, 37). This strongly suggests that HIV-1-specific CD8+ T cells play a role in the control of viral replication. This hypothesis is supported by studies that have documented a relationship between CD8+ T cells and the control of HIV-1 (9, 24) and simian immunodeficiency virus (21, 29, 40) replication. Furthermore, recent studies have shown qualitative differences between the CD8+ T-cell-mediated responses to HIV-1 in ES and those seen for patients with progressive disease (1, 7, 31, 39), and a rebound in viremia is seen after the depletion of CD8+ T cells in monkeys that spontaneously control simian immunodeficiency virus to levels below the limit of detection (16).
We recently provided evidence that CD8+ T cells exert selective pressure on gag in HLA-B*57+ ES (4). We used an ultrasensitive PCR method to amplify and characterize plasma gag sequences from these patients. A comparison of plasma clones to clones amplified from resting CD4+ T cells revealed a striking discordance in the frequencies of mutations in HLA-B*57-restricted epitopes. While rare in clones amplified from resting CD4+ T cells, escape mutations were present in every plasma gag clone, suggesting that HIV-1-specific CD8+ T-cell responses exerted strong selective pressure on this gene (4). It is not known whether other HIV-1 genes are subjected to the same selective pressure. A recent cohort study demonstrated that HIV-1-specific immune responses were focused on Gag in patients who control HIV-1 replication (23). In contrast, responses to accessory and regulatory proteins, including Nef, were associated with higher viral loads (23). Another study found a correlation between a CD8+ T-cell response to the HLA-B*57-restricted Nef epitope HTQGYFPDW (116 to 124) and clinical nonprogression, suggesting that responses to this peptide are protective (35). Interestingly, no correlation between responses to three other HLA-B*57-restricted Nef epitopes and clinical status was seen.
We were thus interested in determining whether selective pressure was exerted on nef in ES. We defined Nef epitopes in nine HLA-B*57-positive ES by enzyme-linked immunospot (ELISPOT) analysis and then amplified nef sequences from resting CD4+ T cells and from free virus from plasma samples from these patients. Sequences from the two compartments were compared to look for evidence of selective pressure. The results provide further evidence of immunological escape from CD8+ T-cell responses in ES.
ES2, ES3, ES5, ES6, ES7, ES8, and ES9 have been previously described (4). ES19 (A*01, A*30, B*07, B*5703) and ES23 (A*03, A*30, B*15, B*5703) first tested positive for HIV-1 in 1996 and 1998, respectively. Both patients have maintained viral loads of <50 copies/ml and CD4 counts of greater than 500 cells/μl since that time. The protocol used was approved by the Institutional Review Board of The Johns Hopkins University. Informed consent was obtained before phlebotomy.
Peripheral blood mononuclear cells (PBMC) from ES were stimulated with overlapping Nef 15-mers obtained from the NIH AIDS Research and Reference Reagent Program. ELISPOT analysis was performed using gamma interferon (IFN-γ)-specific monoclonal antibodies from Mabtech as per their instructions. All plates were evaluated with an automated ELISPOT reader system (Carl Zeiss MicroImaging, Inc.) with KS4.8 software by an independent scientist in a blinded fashion (Zellnet Consulting). Duplicate wells were tested for each peptide, and the mean was determined. Positive responses were defined as peptides that elicited more than 50 spot-forming cells (SFC) per well and had more than three times as many spots as the unstimulated control wells. Unstimulated PBMC uniformly had less than 30 SFC/well, with the exception of PBMC from ES7, which routinely exhibited high background levels. Positive responses in this patient were thus defined as being represented by wells that had five times as many SFC as unstimulated wells. In cases where two adjacent overlapping peptides were both targeted, they were not counted as separate epitopes. The peptide that elicited the stronger response was considered to be the epitope.
To determine whether plasma variants were escape mutants, the corresponding peptides were synthesized either at the Johns Hopkins oncology peptide synthesis facility or at Genemed Synthesis Inc. (San Antonio, TX) and were run in triplicate in the ELISPOT assay at a range from 0.01 to 10 μg/ml using either PBMC or a cell line specific for the proviral variant peptide.
An amplicon spanning full-length env and the majority of nef was amplified from plasma by reverse transcription with the gene-specific primer RT4.2 (GCT CAA CTG GTA CTA GCT TGA AGC ACC) followed by outer PCR with 5envout (ATG GCA GGA AGA AGC GGA GAC AG) and RT4.2 as described previously (2). nef was then amplified by nested PCR with 30 cycles and an annealing temperature of 55°C with 5′ primer “5nefin” (CGT CTA GAA CAT ACC TAG AAG AAT AAG ACA GG) and 3′ primer “RT4.2 shift” (GCT TGA AGC ACC ATC CAA).
Raw nef sequences from plasma and resting CD4+ T cells, along with an M group ancestral sequence, were aligned using Gene Cutter (http://www.hiv.lanl.gov/content/sequence/GENE_CUTTER/cutter.html) and subsequently visually inspected using the sequence editor BioEdit version 7.0.9 (Tom Hall, Ibis Biosciences, Carlsbad, CA). Phylogenies were estimated using both a “classical” approach (41) and a Bayesian approach (20), both functioning under a maximum likelihood optimality criterion. The “classical” approach (41) was implemented using a Web-based version of RAxML (42) available through the CIPRES supercomputing cluster (http://www.phylo.org/). The general time-reversible model of nucleotide substitution with an estimation of the proportion of invariant sites, and gamma-distributed rate variation, were used for the phylogeny estimation, with the M group ancestral sequence used as an outgroup. The precision of phylogenetic reconstruction (nodal support) was assessed via bootstrap analysis, with the number of bootstrap pseudoreplicates determined empirically by the software. The Bayesian approach (20) was implemented using a Web-based version of Mr Bayes, also available through the CIPRES Web portal. Again, the general time-reversible model of nucleotide substitution with an estimation of the proportion of invariant sites, and gamma-distributed rate variation, were used for the phylogeny estimation, with the M group ancestral sequence used as an outgroup. Bayesian inference, coupled with the Markov chain Monte Carlo method, was run for each patient, with 10 chains starting from a random tree. Each chain ran for 2.0 × 107 generations, with samples taken every 100th generation. Phylogenetic trees were visualized using Tree View version 1.6.6 (36).
Plasma and proviral sequences from each patient were translated in BioEdit (http://www.mbio.ncsu.edu/BioEdit/bioedit.html), and the number of amino acid substitutions between the two compartments was calculated by carrying out pairwise comparisons between every plasma and proviral sequence and enumerating the differences. A mean number of substitutions was calculated and used to derive the predicted number of substitutions per site, given the assumption of a uniform distribution of mutations and equal probabilities. This was accomplished by dividing the mean by the total number of amino acid residues. Multiplying this number by 9 gave the predicted number of amino acid substitutions in the epitope where we saw the most evidence of selective pressure (KF9, Nef 82-90). The observed number of substitutions in this region was calculated again by carrying out pairwise comparisons between all plasma and proviral clones and enumerating the differences. A mean of these differences was calculated and used for comparison with the predicted number of substitutions in that region. In order to evaluate statistical significance, a goodness-of-fit test was carried out for each patient, comparing the observed mean differences in that region with the predicted number of differences in that region. This was accomplished by performing likelihood ratio tests of the observed and predicted values for each patient by use of the statistical environment R value.
The nucleotide sequences determined in this study have been submitted to GenBank under accession numbers FJ430356 to FJ430471.
We studied Nef-specific cytotoxic T-lymphocyte responses in nine HLA-B*57 ES by use of overlapping peptides spanning the entire open reading frame. As shown in Table Table1,1, a median of two epitopes (range, zero to five) was targeted in each ES as determined by the IFN-γ ELISPOT assay. Of the total of 21 responses detected for the nine ES, 5 were confirmed to be against one of the four previously defined HLA-B*57-restricted epitopes.
A median of 13 independent nef clones (range, 2 to 18) was amplified from virus in the plasma of five ES. For these five ES, we also amplified a median of 11 nef clones from proviral DNA in resting CD4+ T cells (range, 4 to 24). These resting CD4+ T-cell sequences are referred to hereafter as proviral clones. In patients ES5, ES6, and ES23, for whom we did not have plasma clones, a median of six independent nef clones was amplified from resting CD4+ T cells. We were not able to amplify nef from ES19 due to the limited amount of DNA available.
Figure Figure11 shows phylogenetic analysis for the five ES for whom we amplified both proviral and plasma nef. For each ES, there was little intermingling of clones obtained from the two different compartments. The proviral clones were generally ancestral to the plasma clones, which is consistent with the model that proviral clones represent archived HIV-1 in the latent reservoir and that plasma clones represent a population that has evolved further, perhaps in response to selective pressure from the HIV-1-specific immune response. In four out of the five patients studied (ES2, ES8, ES7, and ES9), most of the sequences observed for the plasma represented a distinct genetic cluster which has undergone further evolution from the clones archived in the latent reservoir.
In order to determine whether the differences between proviral and plasma nef sequences were due to escape mutations in CD8+ T-cell epitopes, sequence analysis was performed. As shown in Fig. Fig.2,2, ES2, ES3, and ES8 all had a pattern of distinct substitutions present in plasma clones but not the proviral clones. These substitutions were located in HLA-B*57-restricted epitope KF9 (Nef 82-90). In this 9-amino-acid epitope, the plasma sequence differed from the proviral sequence by 2 or 3 amino acids for each ES. For example, in the case of ES2, the vast majority of the proviral clones at two different time points (21/24) had the clade B consensus sequence in the KF9 epitope. However, the majority of plasma clones (12/13) had two point mutations in this epitope. ES7 and ES9 also had substitutions in this epitope, but they were the same in clones derived from plasma and resting CD4+ T cells.
There was little evidence of consistent selective pressure present in the other three HLA-B*57-restricted epitopes. For ES9, the H116N escape mutation in the HW9 epitope (Nef 116-124) (13, 35, 38) was present in all clones amplified from both resting CD4+ T cells and plasma. This substitution was also present in the HLA-B*57+ subject who transmitted the virus to her, and thus this mutation most likely was present in the transmitted virus (3). Similarly, the Y120F mutation, located in the overlapping HLA-B*57-restricted epitopes HW9 and YT9 (Nef 120-128), was found in all cellular and plasma clones amplified from ES8. The GF14 epitope (Nef 130-143) contained common substitutions in clones amplified from the two compartments in most ES. Differences between cellular and plasma sequences were seen for just one of the non-HLA-B*57-restricted epitopes, Nef 105-119, which was targeted in ES3.
To determine whether the observed sequence changes in the epitopes occurred by chance alone, we performed a statistical analysis looking at the frequency of changes in the epitope versus the whole Nef protein. For ES2, ES3, and ES8, the presence of two or three mutations in epitope KF9 (Nef 82-90) in plasma compared to provirus suggested that this region may be under selective pressure to evolve at a rate higher than that for the rest of the gene. To provide statistical support for this observation, we performed goodness-of-fit likelihood ratio tests of the observed and predicted numbers of differences in this epitope. The results shown in Table Table22 support the observation that for patients ES2, ES3, and ES8, the numbers of substitutions in the epitope are higher than would be predicted by chance.
To determine whether the substitutions seen in Nef epitopes in the plasma clones represented escape mutations, ELISPOT analysis was performed with the cellular and plasma variants of each epitope. As shown in Fig. Fig.3,3, the proviral variant of Nef 82-90 in ES8 (KSALDLSHF) elicited a stronger IFN-γ response than the plasma variant (TAALDMSHF; mutations are underlined here and below). Thus, the plasma variant is most likely an escape mutant. In ES3, the response to the proviral variant KAALDLSHF was positive at just one concentration (1 μg/ml), whereas no response was seen for the plasma variant QAAHNLSHF at any concentration. ES7 and ES9 had the same substitutions in the KF9 epitope, and in both patients, the same variants were present in plasma and resting CD4+ T cells. Neither subject had a detectable response to their autologous KF9 variants (KGALDLSHF) or to the consensus clade B version of the epitope (Fig. (Fig.33).
We did not have sufficient PBMC to directly test whether the plasma KF9 variant seen in ES2 represented an escape mutant, so a cell line specific for the proviral variant (KAAVDLSHF) was made from PBMC from ES19, a subject who had a strong response to this epitope. As shown in Fig. Fig.4,4, the response to the plasma variant KAAVDISHY was barely detectable, whereas a robust response to the wild-type peptide was seen, suggesting that this plasma variant was in fact an escape mutant in ES2. Given the minimal response to KF9 by ES3, a cell line was also made to the proviral variant (KAALDLSHF) by use of PBMC from ES23. The response to the wild-type proviral variant greatly exceeded the response to the plasma variant (QAAHNLSHF), which again suggested that the plasma variant represented an escape mutant in ES3.
We next looked at the other HLA-B*57-restricted Nef epitopes. The H116N substitution was seen in all the Nef clones amplified from ES9. Prior studies have shown that this variant is an escape mutant (13, 35, 38). We thus compared the CD8+ T-cell response to this peptide as well as to the wild-type epitope. Minimal response was seen to the wild-type peptide at the highest concentration tested, whereas there was no response to the autologous variant (Fig. (Fig.5A).5A). We also compared the overlapping wild-type HW9 and YT9 epitopes to the autologous variants containing the Y120F mutation seen for ES8. As seen in Fig. Fig.5,5, there was a minimal response to both wild-type and plasma variants of both epitopes.
Finally, we looked at the response to the proviral and plasma variants of Nef 105-119 in ES3. A robust response to the proviral variant was seen, whereas almost no response was elicited to the plasma variant which contained the Q107R substitution (Fig. (Fig.5D).5D). It thus appears that the plasma variant represents an escape mutant.
There is now mounting evidence that many ES are infected with replication-competent virus. Full-length sequencing of provirus and plasma virus in large numbers of these patients has not revealed large deletions or signature mutations (33). Furthermore, genotypic and phenotypic analyses of replication-competent virus cultured from the latent reservoir of ES isolates have shown no evidence of attenuation for these isolates (8). We also recently described a patient who maintained an undetectable viral load for a year before he eventually developed virological breakthrough with a viral load of 13,000 copies/ml in the setting of new cytotoxic T-lymphocyte escape mutations (5). Finally, we recently demonstrated the transmission of HIV-1 from a patient who developed full-blown AIDS to an ES who has maintained undetectable viral loads and normal CD4 counts for more than 10 years (3).
The mechanism involved in this remarkable control of fully replication-competent HIV-1 has not been fully defined. The overrepresentation of HLA-B*27 and B*57 (12, 25, 32, 37) suggests that CD8+ T cells may play an active role. It has been shown that Gag is the focus of the immune response in patients who control viral replication (23). Furthermore, in HLA-B*57-positive ES, the Gag-specific CD8+ T-cell response is focused on four HLA-B*57-restricted epitopes (32). Surprisingly, we showed that these ES have escape mutations in at least two of these epitopes in every plasma clone (4). This supported a study showing that some HLA-B*57 ES had escape mutations present in their resting CD4+ T cells (30). It is not known how ES continue to control viral replication in light of these escape mutations. Of note is that while there have been many studies documenting the presence of escape mutations in HIV-1, there are few instances where the development of escape mutations has been temporally associated with a loss of control of viral replication (5, 6, 14, 17). It is possible that this is partially due to a diminished fitness of escape mutants. It has been shown, for example, that the T242N mutation in the HLA-B*57-restricted Gag epitope TW10 exerts a significant cost on fitness (28), and reversion to the wild type occurs when the virus is transmitted to HLA-B*57-negative individuals (27). Compensatory mutations that improve the replicative capacity of this mutant (10) are seen for isolates from HLA-B*57-positive patients with progressive disease but are less common in long-term nonprogressors (LTNP) (34) and ES (10). It thus follows that some escape mutations would have a greater impact on the fitness of HIV-1 in ES than in progressors. It is also possible that residual responses to mutated epitopes (4) or CD8+ T-cell responses to epitopes in other proteins inhibit viral replication.
A recent study suggested that while escape mutations in the HW9 Nef epitope are present in both HLA-B*57 LTNP and progressors, LTNP were more likely to maintain their response to the HW9 Nef epitope (35). Interestingly, only two of the nine ES we studied had strong responses to this epitope. Furthermore, we saw little evidence of escape in this epitope for the five ES from whom we obtained plasma clones. ES9 had the H116N mutation in all clones amplified from both resting CD4+ T cells and plasma. The same mutation was also seen in all clones amplified from the subject who transmitted HIV-1 to her (3). In contrast, we showed earlier that there was discordance between the plasma and proviral clones in an HLA-B*27-restricted Gag epitope in this subject (4). It is thus possible that she was infected with an isolate that contained this mutation and thus that no selective pressure was generated at this site. ES8 also had a mutation (Y120F) in the overlapping HW9 and YT9 epitopes in all plasma and proviral clones as well as in two replication-competent isolates cultured from resting CD4+ T cells (8). This pattern is very different from what we demonstrated for three Gag epitopes (4) and for the KF9 Nef epitope from this patient, where different substitutions were observed in plasma versus proviral clones. It is thus possible that this ES was also infected with an isolate containing this substitution.
It is not clear why the HW9 epitope was not frequently targeted in our cohort and why the H116N escape mutation did not occur more frequently. It is possible that this is due to the relatively small number of patients analyzed in each cohort. Another key difference is that while the LTNP studied were mostly HLA-B*5701 or HLA-B*5801 positive (35), our ES are mostly HLA-B*5703 positive. Finally, ES who control viral loads to <50 copies/ml are likely to have immune responses different from those of LTNP who are viremic.
We did see strong evidence of immune escape in the KF9 epitope in three ES. Interestingly, the two ES who did not have discordance between plasma and provirus in this epitope had the previously described A83G substitution (26). This mutation was present at a higher frequency in HLA-B*5701/B*5801 individuals in clade B (26) and C (22, 26) cohorts, and vertical transmission studies (26, 38) have documented the selection of this variant in HLA-B*57+-infected patients. Functional studies have confirmed that this substitution represents an escape mutation (26). It is likely that ES9 was infected with an isolate containing this escape mutation, as the same substitution was seen in all isolates that were cultured from resting CD4+ T cells from her HLA-B*57+ partner (3). It has been shown that this mutation persists after transmission to HLA-B*57-negative individuals, and this may explain its high frequency in HIV-1 isolates from the general population (26). Given this high frequency, it is possible that ES7 was also infected with an isolate that contained this mutation.
It is interesting that different pathways to immunological escape were chosen by viruses in patients expressing the same allele. The previously described A83G substitution was seen in two ES, but three other ES had novel mutants containing two or three substitutions in the 9-amino-acid epitope. The degree of immunological escape seen with these mutants as determined by ELISPOT analysis appears to exceed what was described for the A83G substitution (26).
Interestingly, while we show here that HLA-B*57-restricted responses do not appear to dominate the Nef-specific immune response in the same way that they do in the case of Gag (32), there appeared to be more immune escape in these epitopes than in the non-HLA-B*57-restricted Nef epitopes. For the five ES for whom plasma and proviral clones were available, sequence discordance was seen in just one non-HLA-B*57-restricted epitope (Nef 105-119 in ES3). In contrast, discordance was seen for HLA-B*57-restricted epitope KF9 in ES2, ES3, and ES8. This possibly could be due to differences in the fitness costs that escape mutations exert on the virus. Furthermore, while both Nef and Gag contain four HLA-B*57-restricted epitopes, plasma/proviral sequence discordance was seen more often for the Gag epitopes (4) than for the Nef epitopes (eight escape mutations in Gag versus three in Nef) for the five ES studied. This may be a reflection of greater ongoing selective pressure exerted on Gag.
While phylogenetic analysis revealed clear differences in plasma versus proviral nef sequences in ES7 and ES9, we did not see differences in targeted epitopes from these patients. It is possible that our ELISPOT screen missed some epitopes from these patients due to the differences in the autologous virus versus the consensus sequence peptides that were used in the screen. An example of this is the KF9 epitope from ES8. While this patient made a robust immune response to his autologous variant of the peptide, no response to the consensus peptide was seen. Thus, our screening strategy would have missed this epitope and the escape mutant at this site. Another possible reason for this phenomenon is that immunological escape had lead to lower IFN-γ responses to both the proviral and plasma variants, and thus some potential epitopes that contain escape mutations were not detected by our screen. The response to KF9 in ES3 is an example of this. An IFN-γ response to proviral variants was seen at only one concentration of peptide (1 μg/ml) which was different from what was used in the screen (5 μg/ml). Thus, the minimal response to the epitope was also missed in our screen.
Overall, our data demonstrate for the first time that CD8+ T-cell-mediated selective pressure is exerted on Nef in HLA-B*57+ ES, especially at the KF9 epitope. The development of escape mutations as a result of selective pressure would be expected in viremic patients, given the low fidelity of reverse transcriptase and the continuous viral replication seen for these individuals. The fact that escape mutants are also common in the plasma of ES who maintain viral loads of <50 copies/ml strongly suggests that there is a low level of ongoing HIV-1 replication taking place.
It remains unclear how HLA-B*57-positive ES are able to maintain viral loads that are similar to those of patients on highly active antiretroviral therapy (11) in spite of immunological escape in both Gag and Nef. It will be interesting to determine whether a similar pattern is seen for HLA-B*57-negative ES. It will also be interesting to determine whether immunological escape occurs in epitopes located in other proteins such as Env, since a robust Env-specific response is associated with higher viral loads (23).
We have recently shown that control in these ES is not due to the recently identified HCP5 single-nucleotide polymorphism (15) that is in linkage disequilibrium with HLA-B*5701(18). We also showed that ES in our cohort have titers of neutralizing antibody to autologous HIV-1 isolates that are lower than those of patients with progressive disease (2); thus, it is unlikely that the humoral immune response can explain the control of viral replication in ES. It is possible that the suppression of viral replication is due to the effect of immune responses to nonmutated epitopes. Support for this hypothesis comes from our finding that immunological escape was not seen for the majority of Nef epitopes identified from these ES. An alternative hypothesis is that while proviral clones with wild-type sequences are fully replication competent, some of the plasma escape mutants are unfit and thus do not replicate well enough to achieve significant levels of viremia. The fact that the plasma KF9 mutants amplified from ES2, ES3, and ES8 have not been previously described in the HIV-1 database may be evidence of diminished fitness. Finally, it is possible that some component of the innate immune response contributes to the control of viral replication. Further studies will be needed to determine how each hypothesis contributes to the elite control of HIV-1 replication seen for these patients.
We thank Jie Xu and Jun Lai for technical assistance.
This work was supported by NIH grant R56 AI73185-01A1 and the Howard Hughes Medical Institute.
Published ahead of print on 22 October 2008.