Cytotoxic T lymphocytes (CTL) play an important role in immune control of acute and chronic viral infections. Analyses of immune responses in individuals infected with human immunodeficiency virus type 1 (HIV-1) indicate that HIV-1-specific CTL are critical in controlling the initial viremia following acute infection and in suppressing viral replication during the chronic phase of infection. Studies of untreated individuals with acute HIV-1 infection show that the decline in viremia in acute infection is associated with the appearance of HIV-1-specific CTL (
6,
7,
22,
38). In animals infected with simian immunodeficiency virus, antibody-mediated depletion of CD8 cells and CTL leads to an increase in viremia, which returns to baseline values when CTL reappear (
31,
48). During chronic infection a negative association between HLA A2 Gag-specific CTL and viral load has been indicated (
42), but other HLA alleles have not been similarly tested. Despite the clear antiviral activity of CTL, these cells fail to eliminate virus and numerous studies indicate that these responses actually decline with disease progression (reviewed in reference
61).
An increasing amount of data has been generated identifying viral peptides that are targeted by CTL, and these peptides largely conform to predicted motifs for HLA binding. CTL recognize infected cells through interaction of the T-cell receptor (TCR) with 8- to 11-amino-acid antigenic peptides complexed with major histocompatibility (MHC) class I molecules. The MHC-peptide complexes arise from intracellular processing of endogenously synthesized viral proteins. Although a large number of peptide epitopes may be generated, T-cell responses are focused to a select number of epitopes, a phenomenon known as immunodominance (reviewed in reference
62). The factors that determine which epitopes will be immunodominant in a given individual have not been clearly defined. Studies of alleles such as B14, B60, and A2 have shown that the majority of persons with these alleles recognize defined epitopes (
10,
25,
36). However, no studies have examined the magnitude and specificity of responses in persons with multiple shared HLA alleles or the relative contributions of specific alleles to the total CTL response.
In this study we analyzed CTL responses in a cohort of subjects matched at the HLA A2, A3, and B7 alleles. These alleles were chosen because they are represented at high frequencies in the population studied (
50), and the HIV-1 epitopes have been well characterized and optimally defined (
9). The breadth, magnitude, and relative immunodominance of HIV-1-specific CTL responses in these individuals to 27 previously defined optimal A2-, A3-, or B7-restricted epitopes was determined by bulk stimulation of peripheral blood mononuclear cells (PBMC) as well as by enzyme-linked immunospot (Elispot) assay and intracellular cytokine staining (ICS). Furthermore, CTL responses to previously defined optimal epitopes restricted by the unshared HLA class I B allele were determined in each subject in order to more completely assess which epitopes are immunodominant in the context of HLA A and B alleles in each individual and to determine the relative contributions of these class I alleles to the total HIV-1-specific CTL response.
MATERIALS AND METHODS
Subjects.
Eight HIV-1-infected subjects were included in this study. Subjects were enrolled through the University of California at San Francisco, the San Francisco City Clinic, and the Massachusetts General Hospital. Subjects 11324, 11841, 13070, 16732, and 221L are individuals with chronic HIV infection receiving highly active antiretroviral therapy (HAART). Subjects AC-03 and OP337 are individuals with acute HIV-1 infection who were treated with HAART within 6 months of seroconversion. Subject 161j is a long-term nonprogressor who has been infected >20 years. The viral loads of the subjects ranged from <50 to 5,390 RNA copies per ml of plasma. All subjects gave written informed consent for these studies. The individual parameters (CD4 count in cells per cubic millimeter/viral load in RNA copies per milliliter of plasma/antiviral therapy) for each of the subjects were as follows: 161j, 1,400/<50/none; 221L, 1,184/478/stavudine, lamivudine, and indinavir; 13070, 358/5,390/zidovudine and lamivudine; 11841, 562/<250/didanosine and stavudine; 11324, 576/697/lamivudine, stavudine, and nelfinavir; 16732, 489/876/zidovudine, saquinavir, and stavudine; AC-03, 947/<50/zidovudine, lamivudine, and nelfinavir; and OP337, 800/1,599/stavudine, didanosine, and nelfinavir.
HLA typing.
HLA typing was performed by the San Francisco Department of Public Health and/or by the Massachusetts General Hospital Tissue Typing Laboratory by standard serological and molecular techniques (
13).
Synthetic peptides.
The 9- to 11-amino-acid peptides used in this study have been reported in the Los Alamos Immunology Database (
9). All were synthesized as COOH-terminal free acids on a Synegy 432A peptide synthesizer (Applied Biosystems, Foster City, Calif.). Lyophilized peptides were reconstituted at 2 mg/ml in sterile distilled water with 10% dimethyl sulfoxide (Sigma) and 1 mM dithiothreitol (Sigma).
Bulk stimulation of peripheral blood mononuclear cells.
Cryopreserved PBMC (4 × 10
6) were stimulated with 10
6 autologous, peptide-pulsed PBMC. PBMC were incubated with each peptide (10 μg/ml) for 1 h at 37°C and then washed three times in R10. Irradiated feeder cells (15 × 10
6allogeneic PBMC) were added to the culture in a 25-cm
2culture flask (Costar, Cambridge, Mass.). Recombinant interleukin-2 (25 U/ml) was added on day 4 and twice a week thereafter. After 10 to 14 days, the cells were assayed on
51Cr-labeled peptide pulsed autologous B-lymphoblastoid cell lines (B-LCL) in a standard
51Cr release assay (
35).
Elispot assay.
Cryopreserved PBMC were thawed in RPMI 1640 medium supplemented with 10% fetal calf serum, 2 mM
l-glutamine, 50 U of penicillin per ml, and 50 μg of streptomycin per ml. Ninety-six-well nitrocellulose plates (Millipore, Bedford, Mass.) were coated with 0.5 μg of human anti-gamma interferon (IFN-γ) monoclonal antibody (MAb) (Mabtech, Stockholm, Sweden) per ml overnight at 4°C. Plates were washed with phosphate-buffered saline (PBS) plus 1% fetal calf serum. PBMC were added at 0.5 × 10
5 and 0.25 × 10
5per well in duplicate wells. For detection of peptide-specific CD8
+ T cells, synthetic peptides (10 μM) corresponding to defined optimal epitopes were added to PBMC. The peptides used are listed in Table
1. Following an overnight incubation at 37°C and 5% CO
2, the plates were washed with PBS before 100 μl of biotinylated anti-IFN-γ MAb (1 μg/ml; Mabtech) was added and incubated at room temperature for 90 min. After the plates were washed again with PBS, 100 μl of 1:2,000-diluted streptavidin-alkaline phosphatase conjugate (Mabtech) was added per well and the plates were incubated at room temperature for 45 min. Wells were washed again with PBS, and individual IFN-γ-producing cells were detected as dark spots after a 30-min color reaction with 5-bromo-4-chloro-3-indolylphosphate and nitroblue tetrazolium using an alkaline phosphatase-conjugated substrate (Bio-Rad Laboratories, Hercules, Calif.). Spots were counted by direct visualization and are expressed as numbers of spot-forming cells (SFC) per 10
6cells. The number of specific IFN-γ-secreting T cells was calculated by subtracting the negative control value from the established SFC count. Wells with greater than 20 SFC/million PBMC were considered positive, based on analysis of seronegative controls (data not shown).
Intracellular IFN-γ staining.
ICS was performed as described previously (
21,
43). Briefly, 1.0 × 10
6 PBMC were incubated with 4 μM peptide and anti-CD28 and anti-CD49d MAbs (1 μg/ml each; Becton Dickinson) at 37°C and 5% CO
2 for 1 h before the addition of GolgiPlug (1 μl/ml; Becton Dickinson). The cells were incubated for an additional 5 h at 37°C and 5% CO
2. PBMC were then washed and stained with surface antibodies, antigen-presenting cell-conjugated anti-CD3 and phycoerythrin-conjugated anti-CD8 (Becton Dickinson) at room temperature for 20 min. Following the washing, the PBMC were fixed and permeabilized (Caltag, Burlingame, Calif.), and the fluorescein isothiocyanate-conjugated anti-IFN-γ MAb (Becton Dickinson) was added. Cells were then washed and analyzed on a FACS-Calibur flow cytometer using CELLQuest software (Becton Dickinson).
DISCUSSION
In this study we have analyzed HIV-1-specific CD8+T-cell responses among a cohort of subjects matched at three common HLA class I alleles. Analysis of these CD8+ responses in eight HIV-1-infected individuals coexpressing the A2, A3, and B7 alleles revealed a wide range in both the breadth and the magnitude of responses. Although all but two of the optimal epitopes reported to be presented by these alleles were recognized by at least one person, the percentages of persons targeting each of the peptides differed over a wide range, as did the magnitude of the responses. The most frequently recognized epitope was A2-restricted p17 77–85, although the A2-restricted response was narrowly directed in most individuals. In addition, the CTL responses to A2-restricted epitopes were not the major contributors to the total HIV-1-specific CTL response by the HLA class I A and B alleles. The patterns of immunodominance also differed significantly among the persons tested, with HLA type not being a predictive factor of which epitopes will be targeted as a dominant response. The lack of recognition of some epitopes suggests that many potential epitopes are not being targeted, and that the immune response to HIV-1 might be broadened in infected persons.
This is the first study, to our knowledge, to analyze the breadth and magnitude of CTL responses to optimal epitopes from several HIV-1 gene products in multiple individuals matched at multiple class I alleles. There has been one previous study, by Goulder et al. (
24), that investigated patterns of immunodominance in two HLA-identical HIV-1-infected siblings. Bulk CTL assays were done with eight optimal epitopes in each sibling and, consistent with our results presented in this paper, the CTL response profile was different for each sibling, in terms of the percent specific lysis for each epitope and which epitopes were targeted as the dominant response. Previous studies by Betts et al. (
5) analyzed HIV-1-specific CTL responses in a cohort of A2-positive individuals and their recognition of the putative immunodominant A2-restricted epitope p17 77–85. Their results agree with ours in that recognition, or lack thereof, of this epitope is not representative of the total HIV-1-specific CTL response. Their study, however, compared individuals matched only at the A2 allele and did not assess similarities or differences of CTL responses to individual optimal A2-restricted epitopes other than the p17 77–85 epitope among the cohort studied.
Several factors likely contribute to immunodominance, including efficiency of processing of peptides, binding of peptides to MHC class I molecules, affinity of T-cell receptors for peptide-MHC complexes, and development of immune escape. It is not clear from the results of this study why individuals of the same HLA type do not target the same epitopes. Immune escape alone cannot account for lack of recognition. Sequencing of autologous virus was performed to begin to address whether lack of response to a particular epitope was due to sequence variation in the autologous virus of the individual. Eight sequences of the B7-restricted Nef epitope 128-137 were obtained from each one of subjects 11841, 16732, and 13070 (data not shown). Subjects 11841 and 13070 both showed an autologous virus sequence identical to the consensus epitope sequence; however, subject 11841 did not recognize this epitope at detectable levels in the Elispot assay, whereas subject 13070 recognized this epitope at 2,980 SFC/million PBMC. Preliminary sequence data therefore indicate that lack of response to a given epitope is not due solely to immune escape and mutation of the epitope. It has been suggested that mutations in the amino acids of the flanking sequences of an epitope may affect efficient processing of the epitope such that it is not presented on the MHC class I allele and potentially resulting in CTL escape (
18,
52,
55). Previous studies have failed to show this in HIV infection (
11). The lack of recognition of specific epitopes thus remains to be explained.
Our data presented here demonstrate a marked degree of heterogeneity in CTL responses, but the degree of heterogeneity is likely to be even larger than shown here because of the expected contribution to the HIV-1-specific immune response of MHC class I C alleles expressed by these individuals, which was not evaluated in this study. Furthermore, this study analyzed CTL responses to epitopes that have been optimally defined for each individual's class I A and B alleles, and it is likely that CTL responses are present to epitopes that have yet to be defined. Studies are under way to define new CTL epitopes in the HIV-1 regulatory proteins, including Tat, Rev, VPR, and Vif (
1,
8). Analysis of CTL responses to these viral proteins has not been included in this study; therefore, the data presented here in terms of the total magnitude and breadth of responses most definitely underestimate the total HIV-1-specific CTL response in these individuals.
It should be noted that many of the subjects were on antiviral therapy at the time of analysis. Although this may result in the diminution in responses over time (
35,
42), established responses are not lost (
3). Even though responses measured may in part be lower in magnitude due to therapy, the study still allows comparative analysis of responses driven by the viral load present in a given individual. Our data also provide additional evidence that the breadth and magnitude of the response can be immense in persons such as 161j who control viremia spontaneously.
In summary, we conclude that there are marked differences in the breadth and magnitude of the CTL response to optimal HIV-1 epitopes in individuals coexpressing the A2, A3, and B7 alleles. These studies thus indicate that HLA type alone is a poor predictor of which epitopes will be targeted in a given individual, and potential epitopes may not be targeted in some persons despite the presence of virus containing peptide sequences predicted to bind to the class I molecule. The identification of immunodominant epitopes targeted in HIV-1 infection has important implications in epitope-based vaccine development. None of the subjects studied targeted all of the potential epitopes that would be predicted to be targeted based on their HLA haplotype. These results indicate that a clear opportunity exists to broaden the repertoire of HIV-1-specific CTL responses and provide rationale for the development of immunotherapy to augment CTL responses in chronic HIV-1 infection.