Abstract
Escape from the CD8+ T cell response through epitope mutations can lead to loss of immune control of HIV replication. Theoretically, escape from CD8+ T cell recognition is less likely when multiple TCRs target individual MHC/peptide complexes, thereby increasing the chance that amino acid changes in the epitope could be tolerated. We studied the CD8+ T cell response to six immunodominant epitopes in five HIV-infected subjects using a novel approach combining peptide stimulation, cell surface cytokine capture, flow cytometric sorting, anchored RT-PCR, and real-time quantitative clonotypic TCR tracking. We found marked variability in the number of clonotypes targeting individual epitopes. One subject recognized a single epitope with six clonotypes, most of which were able to recognize and lyse cells expressing a major epitope variant that arose. Additionally, multiple clonotypes remained expanded during the course of infection, irrespective of epitope variant frequency. Thus, CD8+ T cells comprising multiple TCR clonotypes may expand in vivo in response to individual epitopes, and may increase the ability of the response to recognize virus escape mutants.
Cytotoxic T lymphocytes are unable to mediate full clearance, or adequate control of viral replication in HIV and SIV infections, in part because they select for escape mutations within viral epitopes (1, 2, 3). However, the cellular immune response may suppress the emergence of escape variants by targeting multiple epitopes, thereby reducing the probability of escape mutations at all epitopes occurring simultaneously in a single virus. Indeed, the most rapid emergence of escape from CD8+ T cells occurs during acute infection when fewer epitopes are targeted (1, 2, 4). After acute infection, a CD8+ T cell response is generated that is directed at multiple epitopes, and a broader response is associated with better control of viral replication and slower disease progression (2, 4, 5).
Theoretically, the immune response may inhibit the emergence of escape by targeting individual epitopes with multiple TCR, thereby increasing the likelihood that sequence variants may be recognized. Recent studies have found that CD8+ T cells in HIV, SIV, EBV, and influenza often target a limited range of epitopes, and that the responses to individual epitopes are oligoclonal in composition (6, 7, 8, 9, 10, 11, 12, 13). For HIV, most studies have focused on single or small combinations of epitopes, with the assumption that they represent the total HIV-specific response (14). However, this assumption may lead to inappropriate conclusions about the overall breadth and clonality of the CD8+ T cell response (15, 16). Furthermore, the identification of TCR targeting HIV has generally relied on in vitro selection, cloning, and long-term propagation of cytolytic T cells (8, 17). However, the antiviral potential of CD8+ T cell clones is often, although not always, better associated with cytokine production than cytotoxic activity (18, 19, 20, 21, 22, 23). Therefore, previous studies were potentially biased toward cytotoxic clones with better in vitro growth characteristics, resulting in an incomplete picture of responding clonotypes.
We used cytokine production to identify HIV-specific CD8+ T cells, and to carry out an unbiased analysis of the number of CD8+ T cell epitopes recognized and the number of TCR clonotypes that target immunodominant epitopes. We also examined the influence of sequence variation and antiretroviral therapy on the diversity and frequency of epitope-specific TCR clonotypes over time.
Materials and Methods
Cell stimulation and analysis
Fifteen-mer peptides overlapping by 11 residues corresponding to sequences of HXBc2/Bal chimeric HIV strain (gag, pol, env, and nef) or HIV SF2 strain (tat, rev, vif, vpr, and vpu) were used. Responses were determined by intracellular cytokine staining (ICS)3 as previously described (24). For sorting of unfixed HIV-specific cells, peptide-matrix analysis was used to identify optimal gag 15-mer peptides (24). A total of 5 × 107 PBMC were stimulated with peptide and incubated at 37°C for 1 h at 107 cells/ml, then at 106/ml for 4 h. Cells were washed and incubated with IFN-γ catch reagent (Miltenyi Biotech, Auburn, CA) according to the manufacturer’s instructions, and stained with fluorochrome-conjugated Abs to IFN-γ (Miltenyi Biotech), CD69, and CD8. IFN-γ+CD69+CD8+ cells were sorted by FACS. All studies were approved by the institutions’ Institutional Review Board.
Identification of TCR β-chain (TCRB) sequences and quantitative clonotypic PCR
Sorted peptide-specific cells (5000 from each stimulation) were lysed and mRNA extracted (Oligotex kit, Qiagen, Valencia, CA). Anchored RT-PCR was performed using a modified version of the SMART (switching mechanism at 5′ end of RNA transcript) procedure (25) and a TCRB constant region 3′ primer for the PCR, to obtain TCRB PCR products from the 5′ end to the start of the TCRB constant region. The PCR product was ligated into the pGEMT Easy vector (Promega, Madison, WI) and used to transform Escherichia coli. Colonies were selected, amplified by PCR with M13 primers, and sequenced. PBMC were lysed in proteinase K (Boehringer Mannheim, Indianapolis, IN), and real-time quantitative PCR (qPCR) was performed on 5 μl cell lysate (equivalent to 50,000 cells) with clonotype-specific primers and probes. Probes were labeled with FAM (6′ carboxyflourescene) and QSY7 quencher (MegaBases, Chicago, IL). A standard curve was plotted for each clonotype and template copies were calculated. Plasmid template for the standards was that used in the sequencing of each third complementarity-determining region (CDR3). Samples were analyzed in duplicate. Full details of all primers, components, and cycling temperatures are available upon request.
Chromium release assay
PBMC were cultured with peptide-loaded autologous EBV-transformed B cells, restimulated on day 7, and incubated for a further week. 51Cr-release assay was performed on day 14, with 51Cr-labeled autologous EBV-transformed B cell targets loaded with SLYNTVATL (SL9) or SLYNTIAVL (1 μg/ml). 51Cr release was detected, and specific lysis was determined as: ((experimental release − spontaneous release)/(total release − spontaneous release)) × 100.
Results
Breadth and specificity of total HIV-specific CD8+ T cell responses
Seven HIV-infected individuals were assessed for their total CD8+ T cell responses to HIV by ICS for IFN-γ using overlapping 15-mer peptide pools covering the entire HIV sequence, thus avoiding bias to previously identified epitopes. Their virologic and clinical characteristics around the time of the study are summarized in Table I⇓. Subjects TX4–7 were noncompliant to therapy. Although most subjects mounted their largest response to gag, there were marked differences in the breadth and magnitude of their responses, irrespective of clinical status (Fig. 1⇓A). We examined their responses to the 122 individual overlapping peptides covering HIV gag (to which all subjects mounted a response). Subjects TX4, 6, and 7 responded to a restricted set of gag epitopes, whereas subjects TX1, 2, 3, and 5 mounted broader responses (Fig. 1⇓B). Subject TX7 was notable in that the total CD8+ T cell response to HIV was almost completely directed against one gag epitope.
Responses to HIV peptides in HIV-infected subjects by ICS. A, CD8+ T cell to peptide pools are expressed as the percentage of IFN-γ+CD8+ T cells. B, CD8+ T cell responses to 122 individual peptides covering gag. Each peak represents the response to an individual peptide.
Clinical details of subjects at time of analysis
Identification of CD8+ T cell clones specific for immunodominant HIV epitopes
To determine the clonotypic diversity of HIV-specific responses, we sorted by flow cytometry CD8+ T cells responding to six HIV epitopes, which elicited the highest frequency responses, in five subjects: TX1, 2, 3, 6, and 7 (indicated with an asterisk in Fig. 1⇑). Identification of cells by ICS requires permeabilization, rendering RNA unamenable to analysis. Therefore, we used cell surface Ab-mediated capture of secreted IFN-γ, and CD69 induction. The TCRB CDR3 of the sorted T cells were amplified by anchored RT-PCR to ensure that all HIV-specific TCRB sequences were amplified without bias to particular TCRBV families. Thus, all epitope-specific clonotypes were represented in the PCR product with a relative frequency reflecting that in the original sorted cell population. Table II⇓ shows the sequences and frequencies of the TCRB CDR3 that define the clonotypes responding to dominant gag epitopes in each subject.
CDR3 sequences of HIV gag epitope-specific CD8+ T cells from five subjects, their frequency out of 100 bacterial clones, and their length with respect to the full genomic sequencea
There was heterogeneity in the number of T cell clonotypes elicited to the epitopes. Three of six epitopes elicited monoclonal responses, whereas the others elicited oligo- or polyclonal responses. Such heterogeneity occurred within the response of TX2 to two gag epitopes (Fig. 1⇑); one epitope elicited a response consisting of 15 clonotypes, while the other was monoclonal. We confirmed the frequency and TCRBV subfamily of their epitope-specific clones by flow cytometry with TCRβ chain-specific mAbs (data not shown). These results show that multiple T cell clonotypes can target individual epitopes during HIV infection. Furthermore, there is considerable CDR3 sequence and length diversity among clonotypes targeting a single epitope in a subject. In addition, there was no bias to particular TCRBV families in any of these responses, and no clear correlation between disease state and the number of clonotypes elicited by the epitopes.
Longitudinal analysis of epitope sequence variability
To assess whether multiple CD8+ T cell clonotypes specific for a particular HIV epitope influence the ability of epitope variants to escape a CD8+ T cell response, we studied subject TX7 who recognized only one epitope in gag (SL9) during a 3-year period. This epitope has been shown to resist escape from CD8+ T cells, partially because CD8+ T cells can recognize multiple variants of this epitope (26, 27). Six different TCR clonotypes targeted this epitope in subject TX7. Therefore, we could address whether epitope variants arose and dominated the viral quasispecies, whether the multiple TCR clonotypes could recognize these variants, and whether multiple clonotypes remained expanded in vivo over time.
TX7 was initially treated with antiretroviral therapy and his viral load was suppressed; however, he subsequently became noncompliant, and his viral load rebounded. Plasma and PBMC virus were sequenced at the SL9 epitope region at multiple time points before and after therapy began. Initially, the SL9 sequence SLYNTVATL was present in all viruses sequenced. Then during therapy, three variants emerged, SLYNTVAVL, SLYNTIATL, and SLYNTIAVL, the latter being predominant. None of the amino acid changes were at MHC anchor residues, and all four peptides can bind to HLA-A2, but have markedly variable recognition by different T cell clones (27). Later, the original sequence reemerged and persisted. This could simply be a result of the loss of selective advantage of the variants due to intermittent compliance with therapy (28). However, these particular variant epitopes have been shown to arise in the absence of therapy (27), and thus, they may have failed to escape recognition by the CD8+ T cell response.
Quantitative clonotype PCR of epitope-specific CD8+ T cells
To determine whether the different CD8+ T cell clonotypes showed differential recognition of the variant epitope sequences, we stimulated PBMC from TX7 at 156 wk after therapy initiation with either the original SL9 or the predominant variant epitope, SLYNTIAVL. The frequencies of responding CD8+ T cells were 2.8 and 2.6%, respectively, suggesting that both peptides stimulated a CD8+ T cell response, and that neither epitope escaped immune recognition. We sorted cells responding to each peptide and probed for the presence of the clonotypes by qPCR. We designed CDR3-specific primers and probes for each of the six T cell clonotypes identified in subject TX7. The sensitivity of this assay was one clonotype cell in 100,000 PBMC (data not shown). The specificity of qPCR for HIV-specific clonotypes of TX7 was confirmed as follows: 1) no amplification of clonotype sequences was seen in PBMC from five uninfected adults; 2) no amplification was seen in the PBMC of TX7 when probed for the two predominant HIV-specific clonotypes of TX6; and 3) no HIV-specific clonotypes were detected in sorted CMV-specific CD8+ T cells from TX7 (Fig. 2⇓A).
Responses to original and variant epitopes. A, PBMC from TX7 (96 wk after therapy initiation) and from five healthy donors (nos. 1–5) were probed by qPCR for the six HIV-specific TX7 clones (Table I⇑). CMV-specific CD8+ T cells from TX7 were also probed. PBMC from subjects TX6 and TX7 were probed for two of TX6’s HIV-specific clones. B, Relative frequencies, by qPCR, of five TX7 clones in sorted CD8+ T cells specific for SLYNTVATL (filled bars) and SLYNTIAVL (open bars). C, TX7 PBMC stimulated with different concentrations of SLYNTVATL (•) or SLYNTIAVL (○). The frequency of responding cells by ICS is expressed as the percentage of IFN-γ+CD8+ T cells. D, Percentage of specific lysis of SLYNTVATL- (▪) or SLYNTIAVL-pulsed (▴), or unpulsed (♦) autologous B cells by SLYNTVATL-specific (left) and SLYNTIAVL-specific (right) T cell lines from TX7.
As expected, SL9-specific CD8+ T cells contained all clonotypes (except clone BV6.4, which was no longer detectable by qPCR at this time point, as discussed in the next section). SLYNTIAVL-specific T cells contained the same clonotypes at the same relative frequencies as SL9-specific T cells, except that clone BV17 was undetectable (Fig. 2⇑B). Thus, the TCR expressed by BV17 only recognized SL9, whereas the TCR expressed by the four other clones were more promiscuous and could respond to both SL9 and the variant epitopes. To confirm that this promiscuity was not due to the high peptide concentration used in the assay, we stimulated with suboptimal peptide concentrations. We found that the responses to both epitopes declined, and that the response to the variant peptide was also apparent at lower peptide concentrations (Fig. 2⇑C). Furthermore, short-term lines stimulated with either SL9 or SLYNTIAVL lysed autologous B cells pulsed with either peptide (Fig. 2⇑D), indicating that at least a proportion of the cells mediated cytotoxicity.
Longitudinal analysis of clonotype frequency
To assess changes in the frequencies of the SL9-specific clonotypes with respect to viral load and epitope sequence, we performed clonotypic qPCR on longitudinal samples of the PBMC of TX7. Fig. 3⇓ shows the changes in viral load, epitope sequence, epitope-specific CD8+ T cell frequency by ICS (lower panel), and epitope-specific clonotype frequency by qPCR (upper panel). The magnitude of the CD8+ T cell response to SL9 correlated positively with fluctuations in viral load resulting from nonadherence to therapy, and multiple T cell clonotypes were present at each time point analyzed, although their relative frequencies changed over the 3-year study period. Clone BV6, while initially well represented, steadily decreased in frequency and was undetectable after 96 wk. Notably, clone BV17, which recognized SL9 but not the variant SLYNTIAVL epitope (Fig. 2⇑A) became undetectable 20 wk after therapy initiation, when SLYNTIAVL was transiently predominant, yet subsequently expanded to become the second most frequent clonotype. Thus, the absolute and relative frequencies of the clonotypes changed not only in relation to viral load, but also with respect to each other and to viral epitope variation.
Clonotype frequency in HIV infection. The frequency of six SLYNTVATL-specific clones of TX7 was measured in PBMC >3 years by qPCR. The frequency of each clone as a percentage of PBMC is represented by a different colored bar (top). The arrow shows when HIV-specific clones were identified. Middle, Plasma and PBMC virus sequence. Blue boxes denote SLYNTVATL, red boxes denote SLYNTIAVL, hatched box denotes the variant mix. Up to 20 sequences were analyzed at each time point. Lower, Frequency of SLYNTVATL-specific IFN-γ+CD8+ T cells in TX7 PBMC (filled bars) determined by ICS, and viral load (red circles).
Discussion
In this study we identified, characterized, and quantified HIV-specific CD8+ T cell clones in an unbiased manner using a novel approach that combines in vitro peptide stimulation, cell surface cytokine capture, flow cytometric sorting, anchored TCRB locus RT-PCR, and clonotypic qPCR. This approach obviated skewing of the clonotype population by prolonged stimulation and propagation. Furthermore, we used an anchored RT-PCR step to amplify all TCRB CDR3 of all T cell clones present in the sorted population, without the incomplete coverage of, and bias to, particular TCRBV families conferred by TCRBV-specific PCR primers. This approach requires no prior knowledge of either the specific epitope or the MHC restricting element, and when combined with qPCR, it allows for sensitive and specific quantification of clonotypes at any point, before or after the time at which clonotypes were originally identified. It should be noted that recent papers have demonstrated heterogeneity in cytokine production by HIV-specific T cells and that not all secrete IFN-γ (20, 21, 22, 23). Therefore, we may be underestimating the frequency of responding T cells. However, although not absolutely complete, this type of approach offers a broader analysis than more limited studies performed in the past with limited sets of peptides or Ags.
We found marked heterogeneity in the epitope breadth and magnitude of HIV-specific CD8+ T cell responses, with no apparent relation to disease state, which could arise as a result of attempts by the immune system to control multiple epitope variants (1, 2, 3). Escape of an individual epitope would be more difficult in the presence of multiple specific T cell clonotypes, which could arise de novo and may act to limit virus escape (27, 29, 30), or may arise as a result of recruitment of new clonotypes in response to epitope variants (31, 32). Thus, we observed marked heterogeneity in the clonality of T cells responding to immunodominant HIV gag epitopes. Indeed, one epitope was targeted by 15 TCR clonotypes, indicating that multiple TCR may expand in vivo in response to a single HIV epitope.
The diversity of clonotypes specific for an HIV epitope could influence the appearance and effects of viral escape mutations. In subject TX7, whose CD8+ T cell response to HIV consisted almost entirely of six clonotypes directed against one gag epitope, there was transient expression of two nonsynonymous mutations in that epitope during antiretroviral therapy. Although these mutations can affect T cell recognition (26, 27), this polyclonal response persisted. Furthermore, even though one clonotype did not respond to the mutated epitope, the viral variants failed to escape the CD8+ T cell response of the four remaining clonotypes, which were able to recognize, produce cytokine in response to, and lyse target cells expressing the mutated epitope. It should be noted that our analysis of recognition of the variant epitope depended on the use of peptide-pulsed targets. To rule out aberrant processing and presentation of the variant epitopes, one would ideally perform this analysis using the Ags encoded in minigenes, or vaccinia virus vectors. Although the viral variants that emerged failed to persist as dominant species, genuine viral escape mutants may not always reach fixation because the selective advantage of the variant might be lost (28). This is especially the case in a subject such as TX7, who was intermittently compliant with therapy. However, these particular variant epitopes have been observed in the absence of therapy and so may represent escape from immunologic pressure (27). Although the epitope variants described in this subject are genuine escape mutants, the ultimate effect of mounting a broad CTL response which can recognize those epitopes is to abrogate the immunologic and virologic consequences of such escape. Thus, while a multiclonal CTL response would not affect the probability of viral escape (indeed, it might increase), it does increase the probability of recognition of these escape mutants.
It is of note that when viral load and HIV-specific CD8+ T cell frequencies were high, the frequency of clonotypes detected by qPCR was higher than that of responding T cells detected by ICS (Fig. 3⇑). Yet when virus was suppressed, the frequencies were equal and the ratio was one. During any time interval, it is likely that only a fraction of all epitope-specific CD8+ T cells will produce an optimal intracellular response, even when exposed to optimal levels of Ag and costimulation (33). Thus, while ICS may only detect those CD8+ T cells which can respond under the specific conditions of the assay, qPCR may detect all clonotype cells, irrespective of whether they are in a state of responsiveness to peptide stimulation.
Our findings indicate that an understanding of HIV epitope escape from CD8+ T cells must take into account the plasticity of the response against particular epitopes. Recognition of one or multiple epitopes, use of multiple TCR clonotypes, and HLA associations may all play roles in controlling HIV replication. Furthermore, in subjects immunized with HIV vaccines, it is possible that the infecting virus would rapidly mutate away from the epitope sequences used to elicit immunity. Our data suggest that strategies which elicit multiple T cell clonotypes against HIV epitopes may potentially provide immunity that can tolerate such mutations.
Footnotes
-
↵1 This work was supported by AMFAR Grant no. 02680-28-RGV (to D.C.D.).
-
↵2 Address correspondence and reprint requests to Dr. Daniel C. Douek, Vaccine Research Center, National Institute of Allergy and Infectious Diseases, National Institutes of Health, Room 3509, 40 Convent Drive, Bethesda, MD 20892. E-mail address: ddouek{at}mail.nih.gov
-
↵3 Abbreviations used in this paper: ICS, intracellular cytokine staining; qPCR, quantitative polymerase chain reaction; CDR3, third complementarity-determining region; TCRB, TCR β-chain.
- Received November 13, 2001.
- Accepted January 8, 2002.
- Copyright © 2002 by The American Association of Immunologists